Method and system for comprehensive preparation and performance guarantee of high-thermal-conductivity epoxy resin-based dry-type transformer and storage medium

Abstract

This application discloses a comprehensive preparation and performance assurance method, system, and storage medium for a high thermal conductivity epoxy resin-based dry-type transformer, belonging to the interdisciplinary field of polymer materials and transformers. The method includes: first, adjusting the performance of the high thermal conductivity epoxy resin composite material according to preset material performance targets to obtain optimized material design parameters; then, testing and evaluating samples prepared according to these parameters, and further revising the material design parameters based on the evaluation results; next, based on the revised parameters, conducting multi-scale modeling, structural design, multi-physics coupled simulation verification, and multi-objective collaborative optimization of the dry-type transformer to obtain optimized global design parameters; finally, fabricating a transformer prototype based on these parameters and verifying its comprehensive performance through experimental methods. This application enables dry-type transformers to achieve efficient heat dissipation and reliable insulation, adapting to the complex operating conditions of renewable energy sources and ensuring stable operation of the equipment throughout its entire life cycle.

Description

Technical Field

[0001] This invention relates to the field of polymer materials and transformer technology, specifically to a comprehensive preparation and performance assurance method, system and storage medium for a high thermal conductivity epoxy resin-based dry-type transformer, which is applied to renewable energy power plants. The invention includes a technical system for cross-scale performance regulation of high thermal conductivity epoxy resin materials, multi-physics field collaborative design of dry-type transformers, prototype preparation, defect detection and closed-loop control of the entire process. Background Technology

[0002] The global energy structure is rapidly transitioning towards a green and low-carbon model, with large-scale grid connection of renewable energy becoming a major trend. This poses severe challenges to the reliability, capacity adaptability, and cost control capabilities of dry-type transformers under complex operating conditions. Dry-type transformers, with their fire-retardant and maintenance-free characteristics, are widely used in critical scenarios such as photovoltaic power plants and wind farms. However, large-capacity products have long relied on imported high-performance epoxy resin raw materials, and core technologies are controlled by international companies, restricting the industry's independent development and cost reduction potential. The high proportion of renewable energy integration leads to increased power fluctuations and a significant increase in harmonic currents, placing dry-type transformers in a complex operating environment with deep coupling of electric, magnetic, temperature, and stress fields.

[0003] Existing technologies have significant shortcomings at the material level. Conventional epoxy resins generally have a thermal conductivity below 0.6 W / mK and lack a multi-scale coordinated control mechanism at the macro, meso, and micro levels, making it difficult to simultaneously meet the precise matching requirements of thermal conductivity, insulation, and mechanical properties. At the design and manufacturing level, the use of a single-physics-field independent verification mode cannot effectively simulate the multi-field coupling effect of electromagnetism, thermodynamics, and mechanics. The heat dissipation structure design is simplistic, and the process is prone to problems such as residual bubbles and structural inhomogeneity. At the detection and verification level, the accuracy of microcrack defect identification is insufficient, there is a significant deviation between accelerated aging tests and actual operating conditions, and a closed-loop feedback system for design parameters, manufacturing processes, and performance verification has not been established, making it difficult for products to adapt to dynamic load fluctuations and harmonic impacts in renewable energy scenarios.

[0004] Existing technological bottlenecks severely hinder dry-type transformers from achieving efficient heat dissipation, reliable insulation, and stable operation throughout their entire life cycle. There is an urgent need to develop technical solutions that integrate precise control of material properties, collaborative design of multiple physical fields, and closed-loop control throughout the entire process. Summary of the Invention

[0005] The purpose of this application is to provide a comprehensive method for the preparation and performance assurance of a high thermal conductivity epoxy resin-based dry-type transformer, which has the ability to dissipate heat efficiently, provide reliable insulation, and operate stably throughout its entire life cycle, adapting to dynamic load fluctuations and harmonic impacts in renewable energy access scenarios.

[0006] Firstly, this application provides a comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer, comprising the following steps: S1. A performance control method based on high thermal conductivity epoxy resin composite material is adopted. The performance of high thermal conductivity epoxy resin composite material is controlled according to the preset material performance target value to obtain optimized material design parameters. The material design parameters include epoxy resin matrix parameters, compound thermally conductive filler parameters, and interface modifier parameters. S2. Using a detection and evaluation method based on high thermal conductivity epoxy resin composite materials, the high thermal conductivity epoxy resin composite material samples prepared according to the material design parameters are detected and evaluated to obtain the sample evaluation results. Based on the sample evaluation results, the material design parameters are corrected to obtain the corrected material design parameters. S3. Using the high thermal conductivity epoxy resin dry transformer design and verification method, based on the modified material design parameters, the dry transformer undergoes multi-scale modeling, structural design, multi-physics field coupling simulation verification, and multi-objective collaborative optimization to obtain the optimized global design parameters of the dry transformer. S4. Using the transformer prototype manufacturing method, a high thermal conductivity epoxy resin-based dry-type transformer prototype is manufactured according to the optimized global design parameters of the dry-type transformer. The transformer prototype is then tested and verified using the transformer prototype testing method to complete the comprehensive preparation and performance assurance of the high thermal conductivity epoxy resin-based dry-type transformer. The performance control method is used to determine the filler combination, interface modifier, and related interface and structural parameters of the high thermal conductivity epoxy resin composite material, so as to achieve precise control of the macroscopic thermal conductivity, electrical strength, and flexural strength of the material; the detection and evaluation method is used to quantitatively detect the defect type, location, and size of the composite material sample, and evaluate the aging degree and remaining life of the sample; the dry-type transformer design and verification method is used to complete the microscopic, mesoscopic, and macroscopic multi-scale modeling of the dry-type transformer, as well as the structural design and multi-physics field coupling simulation verification of the dry-type transformer based on the renewable energy access scenario; the transformer prototype manufacturing method is used to complete the molding and overall casting and curing of the transformer coil and core; and the transformer prototype testing method is used to verify the inter-turn insulation, thermal conductivity, and mechanical properties of the transformer prototype.

[0007] Further, in step S1, based on the preset performance target value, a model parameter set is obtained by screening filler combination schemes and corresponding interface modifiers through high-throughput calculations. Specifically: Based on a pre-set filler database, a performance prediction proxy model is trained using a machine learning algorithm. The inputs of the performance prediction proxy model include filler type, filler shape, filler volume fraction, and molecular descriptor of the interface modifier. Using the performance target value as the optimization objective, a multi-objective optimization algorithm is used to perform a global search in the prediction space of the performance prediction proxy model to select the Pareto optimal solution set. By extracting common features from the Pareto optimal solution set, the filler combination scheme and the corresponding interface modifier are determined, forming a model parameter set.

[0008] Furthermore, the step of determining the filler combination scheme and corresponding interface modifier by extracting common features from the Pareto optimal solution set to form a model parameter set includes: The Pareto optimal solution set is grouped using a clustering algorithm. The characteristic variables of the clustering algorithm include filler type, filler shape, filler gradation ratio, and molecular descriptor of interface modifier. Based on each characteristic variable, a characteristic statistical matrix is ​​established to quantitatively calculate the average aspect ratio of the filler, the variance of the aspect ratio distribution, the gradation ratio of the filler, and the numerical concentration interval of the molecular descriptor of the interface modifier. Based on the feature statistics matrix, principal component analysis is used to identify the feature combinations that affect the performance of high thermal conductivity epoxy resin composites, and filler combination schemes that appear repeatedly in multiple Pareto optimal solutions and whose frequency exceeds a preset threshold are selected. The structural parameters of the interface modifier most strongly correlated with the filler combination scheme were determined by correlation analysis. The structural parameters include molecular chain length, type and number of polar groups, and molecular conformation characteristics. A model parameter set is constructed based on the selected filler combination schemes and the structural parameters of their corresponding interface modifiers.

[0009] Furthermore, in step S1, the optimal interface parameters are obtained by calculating the interface binding energy and interface thermal conductivity through molecular dynamics simulation based on the model parameter set. Specifically: Based on the structural parameters of each filler combination scheme and its corresponding interface modifier in the model parameter set, a three-component atomic model including epoxy resin matrix, filler surface and interface modifier is constructed. Energy minimization and molecular dynamics relaxation simulations were performed sequentially on each three-component atomic model to obtain a stable equilibrium system structure. Based on the aforementioned equilibrium system structure, the interfacial binding energy corresponding to each filler combination scheme is calculated using energy analysis methods in molecular dynamics simulations. The interfacial binding energy is used to characterize the adsorption strength between the interfacial modifier and the filler surface. Based on the non-equilibrium molecular dynamics method, a temperature gradient is established at both ends of the equilibrium system structure, and the interfacial thermal conductivity corresponding to each packing combination scheme is calculated by the steady-state heat flow method. Based on the calculation results of interfacial binding energy and interfacial thermal conductivity corresponding to all filler combination schemes, quantitative correlation models between the molecular structure parameters of the interfacial modifier and the interfacial binding energy and interfacial thermal conductivity were established by multiple regression analysis. Based on the quantitative correlation model, a multi-objective optimization algorithm is used to iteratively optimize the molecular structure parameters of the interface modifier with the goal of simultaneously maximizing the interfacial binding energy and interfacial thermal conductivity, so as to obtain the optimal interfacial parameters.

[0010] Furthermore, based on the calculated interfacial binding energy and interfacial thermal conductivity of all filler combination schemes, quantitative correlation models between the molecular structure parameters of the interfacial modifier and the interfacial binding energy and interfacial thermal conductivity are established through multiple regression analysis, including: The molecular structure parameters of the interface modifier are used as the set of independent variables, and the interface binding energy and interface thermal conductivity are used as the dependent variables, respectively, to construct an initial multivariate regression dataset. The initial multivariate regression dataset is standardized, and the variance inflation factor method is used to diagnose multicollinearity. Independent variables with variance inflation factors greater than a preset threshold are removed to obtain the processed dataset. Based on the processed dataset, the stepwise regression method was used to screen the molecular structure parameters of the interface modifiers with the p-value of the statistical test being less than a preset threshold, and an initial quantitative correlation model was established. The prediction accuracy of the initial quantitative correlation model was evaluated using k-fold cross-validation. Based on the evaluation results, the coefficients of the initial quantitative correlation model were regularized and optimized using ridge regression algorithm to obtain a quantitative correlation model between the molecular structure parameters of the interface modifier and the interface binding energy and interface thermal conductivity.

[0011] Furthermore, in step S1, based on the optimal interface parameters, the mesoscopic structural characteristic parameters are obtained by simulating the packing dispersion process and interface layer evolution process using the phase-field method. Specifically: Based on the optimal interface parameters, a three-phase phase field model is established, which includes an epoxy resin matrix phase, a filler phase, and an interface modifier phase. Three sets of sequence parameters are defined to characterize the distribution state of the epoxy resin matrix phase, the filler phase, and the interface modifier phase, respectively. Based on the interface bonding energy and interface thermal conductivity in the optimal interface parameters, the interface energy parameters and gradient energy coefficients between each phase in the three-phase phase field model are set. Run the three-phase phase field model until the system reaches a dynamic equilibrium state; The spatial distribution, orientation distribution, packing network shape, and interface layer thickness distribution of the packing under dynamic equilibrium conditions are extracted as simulation results. Based on the simulation results, the permeation threshold, thermal conductivity path density, average thickness of the interface layer and its distribution uniformity of the packing network are quantitatively calculated to form a set of mesoscopic structural feature parameters that include packing distribution characteristics and interface layer characteristics.

[0012] Furthermore, running the three-phase phase-field model until the system reaches a dynamic equilibrium state includes: In the three-phase phase field model, a convection term describing the shear flow field is introduced. By coupling the flow field control equation and the phase field evolution equation, the dispersion process, migration process and agglomeration kinetics of the filler in the epoxy resin matrix are simulated until the system reaches a dynamic equilibrium state. Based on the optimal interface parameters, the interface gradient parameters are set, and the adsorption process of the interface modifier on the filler surface and the formation and evolution process of the interface layer are simulated by solving the interface evolution equation until the system reaches a dynamic equilibrium state.

[0013] Furthermore, the convection term describing the shear flow field is introduced into the three-phase phase field model. By coupling the flow field control equation and the phase field evolution equation, the dispersion, migration, and aggregation dynamics of the filler in the epoxy resin matrix are simulated until the system reaches a dynamic equilibrium state, including: Based on the actual processing parameters, the flow field parameters in the three-phase phase field model are initialized, including the shear rate, viscosity and density of the epoxy resin matrix; Establish the coupled flow field control equations and phase field evolution equations; A convection term describing the shear flow field is introduced into the phase field evolution equation. The convection term is composed of the product of the flow field velocity vector and the order parameter gradient, and is used to describe the influence of the flow field on the packing transport and interface evolution. The coupled equations are spatially discretized using the finite difference method, and time integration is performed using an implicit-explicit hybrid time-progression method. Within each time step, the flow field control equations are first solved to obtain the updated flow field velocity distribution. Then, the flow field velocity distribution is substituted as a known quantity into the phase field evolution equations to solve for the order parameters of each time step. Based on the sequence parameters at each time step, the dispersion, migration and aggregation kinetics of fillers in the epoxy resin matrix are simulated by an iterative method until the system reaches a dynamic equilibrium state. The process of setting interface gradient parameters based on the optimal interface parameters, simulating the adsorption process of the interface modifier on the filler surface and the formation and evolution of the interface layer by solving the interface evolution equation, until the system reaches a dynamic equilibrium state, includes: Based on the interfacial binding energy in the optimal interfacial parameters, the adsorption energy parameters of the interfacial modifier on the filler surface are determined, and the interfacial gradient energy coefficient in the three-phase phase field model is set based on the adsorption energy parameters. Initialize the concentration distribution field of the interface modifier in the epoxy resin matrix and set adsorption boundary conditions on the filler surface; A non-conservative Allen-Cahn equation describing the evolution of the interface layer is established. The non-conservative Allen-Cahn equation includes an interface gradient term, a double-well potential function term, and a convection transport term. Based on the interfacial gradient energy coefficient and the adsorption energy parameter, the interfacial evolution equation is solved to simulate the adsorption process of the interfacial modifier on the filler surface and the formation and evolution process of the interfacial layer until the system reaches a dynamic equilibrium state.

[0014] Further, in step S1, the macroscopic thermal conductivity, electrical strength, and flexural strength of the high thermal conductivity epoxy resin composite material are predicted by finite element analysis based on the mesoscopic structural characteristic parameters, resulting in predicted performance values. Specifically: Based on the filler spatial distribution, orientation distribution, and filler network shape in the mesoscopic structural characteristic parameters, a three-dimensional geometric model of the high thermal conductivity epoxy resin composite material is reconstructed. Based on the filler type and the thickness distribution of the interface layer, the filler phase, the interface layer and the epoxy resin matrix phase are respectively assigned corresponding material properties, including thermal conductivity, dielectric constant, electrical conductivity and elastic modulus; The three-dimensional geometric model is meshed, and finite element models of the heat conduction control equation, the electrostatic conduction control equation, and the linear elasticity control equation are established respectively. Temperature boundary conditions are applied to the finite element model of heat conduction, and the temperature field distribution is obtained by solving. The macroscopic thermal conductivity of the high thermal conductivity epoxy resin composite material is calculated based on Fourier's law. Voltage boundary conditions are applied to the finite element model of electrostatic conduction, and the electric field distribution is obtained by solving. The electrical strength of the high thermal conductivity epoxy resin composite material is predicted based on the extreme values ​​of the electric field strength. Displacement and stress boundary conditions are applied in the finite element model of online elasticity, and the stress-strain field is obtained by solving. The bending strength of high thermal conductivity epoxy resin composite material is predicted based on the maximum stress criterion. The macroscopic thermal conductivity, electrical strength, and flexural strength are used as performance prediction values.

[0015] Further, in step S1, the performance is controlled by adjusting the filler volume fraction and interface modification strength based on the difference between the predicted performance value and the target performance value. Specifically: Calculate the relative error between the predicted performance value and the target performance value, wherein the predicted performance value includes the predicted macroscopic thermal conductivity, the predicted electrical strength, and the predicted flexural strength; A multi-objective optimization function is established based on the relative error, with the optimization objective being to minimize the weighted sum of the relative errors of the predicted macroscopic thermal conductivity, predicted electrical strength, and predicted flexural strength. Key control parameters for macroscopic thermal conductivity, electrical strength and flexural strength are identified. These key control parameters include filler volume fraction and interface modification strength, wherein the interface modification strength is characterized by the functional group density and molecular chain length of the interface modifier. Based on gradient descent or genetic algorithm, the filler volume fraction and interface modification intensity are iteratively optimized within a preset parameter adjustment range, wherein the interface modification intensity is characterized by the functional group density and molecular chain length of the interface modifier. When the value of the multi-objective optimization function converges within the preset tolerance range, or reaches the maximum number of iterations, the optimization process is stopped, and the optimal combination of filler volume fraction and interface modification strength is obtained. Performance is controlled based on the optimal combination of filler volume fraction and interfacial modification strength.

[0016] Further, in step S2, based on the application conditions and expected performance indicators of the high thermal conductivity epoxy resin composite material in dry-type transformers, a calibration sample containing microcracks of a predetermined morphology is prepared, specifically: Based on the application conditions and expected performance indicators, the preset morphological parameters of microcracks are determined through simulation analysis. The expected performance indicators include thermal conductivity, volume resistivity, flexural strength and heat deformation temperature. The preset morphological parameters include the number of microcracks, the length of microcracks, the width of microcracks, the depth of microcracks and their spatial distribution coordinates. The calibration sample to be prepared is fixed according to its actual assembly orientation in the dry-type transformer winding; By applying controlled mechanical stress or thermal shock treatment, microcracks consistent with the preset morphological parameters are introduced into the fixed calibration sample to obtain the calibration sample containing microcracks.

[0017] Further, in step S2, the calibration sample is scanned with multi-frequency electromagnetic waves to acquire and process the electromagnetic wave response signal, generating the electromagnetic wave response characteristic spectrum of the calibration sample. Specifically: Based on the material properties of the calibration sample and the preset morphological parameters of the microcracks, the scanning parameters of the multi-frequency electromagnetic waves are set. The material properties of the calibration samples include thermal conductivity, dielectric constant, and elastic modulus. The scanning parameters include a frequency range of 0.1THz to 2.7THz and a scanning resolution of not less than 6GHz; Based on the scanning parameters, a full-coverage scan is performed on the calibration sample fixed in the actual assembly position of the dry-type transformer winding, and the original electromagnetic wave response signal is acquired in real time. The original electromagnetic wave response signal is preprocessed to obtain a purified signal. The preprocessing includes wavelet threshold noise reduction and matched filtering enhancement. Extract characteristic parameters, including characteristic frequency, characteristic amplitude, reflection coefficient, and projection coefficient, from the purified signal; The characteristic parameters are integrated and normalized to generate the electromagnetic wave response characteristic spectrum of the calibration sample.

[0018] Further, the step of integrating and normalizing the characteristic parameters to generate the electromagnetic response characteristic spectrum of the calibration sample includes: Based on the characteristic frequency and the characteristic amplitude, a clustering analysis method is used to perform pattern recognition to obtain a cluster of characteristic parameters representing different microcrack types; Based on the cluster of characteristic parameters, a multi-level micro-defect information superposition algorithm is used to perform spatial domain fusion processing on the reflection coefficient and projection coefficient to obtain enhanced defect spatial distribution characteristics. Based on the cluster of characteristic parameters and the enhanced spatial distribution characteristics of defects, a quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters is constructed. Based on the quantitative mapping relationship, the characteristic parameters of the entire frequency band are normalized and encoded to generate an electromagnetic wave response characteristic spectrum containing characteristic frequency bands, characteristic amplitudes, and spatial distribution coordinates.

[0019] Further, the step of using a multi-level micro-defect information overlay algorithm to perform spatial domain fusion processing on the reflection coefficient and projection coefficient based on the feature parameter cluster to obtain enhanced defect spatial distribution features includes: Based on the cluster of characteristic parameters, the amplitude gradient distribution of the reflection coefficient and the phase delay distribution of the projection coefficient are adaptively divided into multiple data subsets according to the scanning space coordinates. The number of the subsets is positively correlated with the density of the microcrack spatial distribution coordinates. Based on the multiple hierarchical data subsets, the local amplitude maxima of the reflection coefficients and the average phase delay of the projection coefficients within each hierarchical level are extracted to construct the feature vector of the current level. Principal component analysis was used to reduce the dimensionality of the feature vectors at each level and fuse them to obtain the defect probability distribution map corresponding to each level. Based on spatial coordinates, the defect probability distribution maps corresponding to each level are weighted and superimposed to obtain the superimposed defect probability distribution map. The superimposed defect probability distribution map is processed by morphological opening operation to filter out noise and connect adjacent defect regions to obtain enhanced defect spatial distribution features.

[0020] Further, in step S2, based on the electromagnetic wave response characteristic spectrum, a clustering analysis and multi-level micro-defect information superposition algorithm are used to extract the quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters, and an electromagnetic feature fingerprint database is established based on the quantitative mapping relationship. Specifically: Based on the electromagnetic wave response characteristic spectrum, the K-means++ clustering algorithm is used to perform unsupervised classification of the characteristic frequency bands and characteristic amplitudes, resulting in an initial mapping relationship lookup table with the cluster center as the index and associated with the mean length and mean width of the corresponding microcracks. Based on the spatial distribution coordinates in the electromagnetic wave response characteristic spectrum, a multi-level micro-defect information superposition algorithm is used to extract the spatial distribution characteristics of defects. Based on the aforementioned spatial distribution characteristics of defects, the correlation features between the depth of microcracks and their spatial distribution coordinates are extracted by Gaussian mixture model clustering. The number of components in the Gaussian mixture model clustering is adaptively determined according to the number of microcracks in the preset morphological parameters. Based on the initial mapping lookup table and the associated features, a microcrack morphology-electromagnetic feature parameter mapping relationship matrix is ​​constructed, which uses feature frequency and feature amplitude as parameters and associates microcrack length, microcrack width, microcrack depth and spatial distribution coordinates. Based on the microcrack morphology-electromagnetic feature parameter mapping matrix, the feature parameters in the electromagnetic wave response feature spectrum are normalized and encoded to establish the electromagnetic feature fingerprint database.

[0021] Furthermore, the step of extracting the correlation features between the depth of microcracks and their spatial distribution coordinates through Gaussian mixture model clustering based on the spatial distribution characteristics of the defects includes: The spatial distribution characteristics of the defects are used as the input feature vector for Gaussian mixture model clustering; The number of components for Gaussian mixture model clustering is adaptively determined based on the number of microcracks in the preset morphological parameters. The expectation-maximization algorithm is used to estimate the parameters and iteratively optimize the Gaussian mixture model until the Gaussian mixture model converges. Based on the converged Gaussian mixture model, the posterior probability of each microcrack data point belonging to each Gaussian component is calculated. Based on the maximum a posteriori probability principle, each microcrack data point is assigned to the corresponding Gaussian component to complete defect clustering; Extract the mean vector and covariance matrix of each Gaussian component, where the mean vector represents the typical depth and spatial coordinates of the current type of microcrack, and the covariance matrix represents the distribution pattern of the typical depth and spatial coordinates of the current type of microcrack. Based on the mean vector and covariance matrix of each Gaussian component, a three-dimensional correlation feature matrix between the depth of the microcrack and its spatial distribution coordinates is constructed, which serves as the correlation feature between the depth of the microcrack and its spatial distribution coordinates.

[0022] Furthermore, in step S2, during the accelerated aging test simulating the operating conditions of a dry-type transformer, the high thermal conductivity epoxy resin composite material sample undergoes multi-stage electromagnetic wave scanning, and electromagnetic wave scanning data is generated by encoding according to a time sequence. Specifically: The high thermal conductivity epoxy resin composite material sample was fixed in the aging test device according to the actual assembly orientation of the dry transformer winding. Based on the actual operating conditions of the dry-type transformer, the environmental parameters for the aging test are set, including: a temperature cycling range of -40°C to +120°C, a relative humidity range of 20% to 95%, a mechanical vibration frequency of 10Hz to 200Hz and an amplitude of 0.5g to 5g, and a voltage level of 0.5 times to 1.5 times the rated voltage. Accelerated aging tests are conducted according to a preset aging cycle, which includes multiple consecutive aging stages. Each aging stage includes a heating sub-stage, a heat preservation sub-stage, a cooling sub-stage, and a voltage loading sub-stage. After each aging stage, the high thermal conductivity epoxy resin composite material sample is scanned using the multi-frequency electromagnetic wave scanning parameters to collect the original electromagnetic wave response signal. The collected raw electromagnetic wave response signal is preprocessed to obtain a purified signal. The preprocessing includes wavelet threshold denoising and matched filtering enhancement. From the purified signal, feature parameters including characteristic frequency, characteristic amplitude, reflection coefficient and projection coefficient are extracted to obtain the feature parameters of each stage; According to the aging time series, the characteristic parameters of each stage are encoded in time series to generate electromagnetic wave scanning data.

[0023] Furthermore, in step S2, a deep learning model is used to detect and evaluate the electromagnetic feature fingerprint database and electromagnetic wave scanning data to obtain quantitative detection results and quantitative evaluation results, specifically: Construct a deep learning model that includes convolutional neural network branches and long short-term memory network branches; The electromagnetic fingerprint database is input into the branch of the convolutional neural network to extract spatial features and obtain static fingerprint features. The electromagnetic wave scanning data is input into the long short-term memory network branch to extract dynamic temporal features. A cross-modal attention fusion module is used to adaptively weight and fuse static fingerprint features and dynamic temporal features to obtain a fused feature vector; The fused feature vector is input into a fully connected classification layer to classify the defect type and regress the spatial location, and output the type and location of defects in the high thermal conductivity epoxy resin composite sample. The fused feature vectors are simultaneously input into a regression layer to quantitatively predict the defect size and output the size of the defect in the high thermal conductivity epoxy resin composite sample. The dynamic time series characteristics are input into a time series analysis layer to evaluate the aging degree of high thermal conductivity epoxy resin composite material samples and predict the remaining life. The predicted values ​​of aging degree and remaining life of high thermal conductivity epoxy resin composite material samples are output. The quantitative detection results are obtained by integrating the outputs of the classification layer and the regression layer. The quantitative evaluation results are obtained by integrating the output of the time series analysis layer.

[0024] Furthermore, step S2 also includes: generating a health status report for the high thermal conductivity epoxy resin composite material based on the quantitative detection results and the quantitative assessment results, wherein the health status report includes: A planar schematic diagram of a high thermal conductivity epoxy resin composite sample, including the type, location, and size of defects. Aging degree of high thermal conductivity epoxy resin composite material sample as a function of time; Predicted remaining life of high thermal conductivity epoxy resin composite material specimens.

[0025] Further, in step S3, microscopic modeling, mesoscopic modeling, and macroscopic modeling are performed sequentially according to the material design parameters to obtain the formulation and structure of the high thermal conductivity epoxy resin. Specifically: Based on the material design parameters, the composite material component system is pre-designed to determine the basic formulation data, including the epoxy resin matrix type, the compound thermally conductive filler system, the interface modifier and its mass fraction range. Microscale molecular dynamics modeling was performed based on the basic formulation data to obtain microscale interface parameters including interfacial binding energy, interfacial thermal conductivity, matrix glass transition temperature, elastic modulus, and dielectric constant. Based on the microscopic interface parameters, a mesoscopic-scale phase-field method model is used to simulate the evolution of the material structure and obtain mesoscopic structural characteristic parameters including filler dispersion uniformity, interface layer thickness, aggregate size and distribution density. Based on the mesoscopic structural characteristic parameters, macroscopic-scale finite element modeling and multiphysics coupling analysis are performed to predict macroscopic performance parameters including thermal conductivity, bending strength and electrical strength. The macroscopic performance parameters are compared with the preset target performance, and the microscopic interface parameters, mesoscopic structure characteristic parameters and filler volume fraction are adjusted by orthogonal iteration method to perform parameter iterative optimization until the difference from the preset target performance is within the preset range, and the high thermal conductivity epoxy resin formulation and structure are output. The microscale molecular dynamics modeling includes analyzing the glass transition temperature of the matrix, the mesoscale phase field modeling includes controlling the evolution of material structure, and the macroscale finite element modeling includes multiphysics coupling analysis.

[0026] Further, in step S3, a high thermal conductivity epoxy resin composite material sample is prepared according to the stated formula and structure and subjected to terahertz detection. The sample evaluation result is obtained through deep learning algorithm analysis. Specifically: A high thermal conductivity epoxy resin composite material sample containing microcracks was prepared according to the aforementioned high thermal conductivity epoxy resin formulation and structure. Wideband terahertz time-domain spectroscopy was performed on the high thermal conductivity epoxy resin composite material sample. By optimizing the photoconductive antenna structure and signal processing algorithm, detection data with a bandwidth of 0.1-2.7THz, a resolution of 6GHz, and a dynamic range of >60dB were obtained. Based on the detection data, a defect fingerprint database containing electromagnetic wave characteristic parameters is constructed, including the reflection coefficient and the transmission coefficient. Based on the aging test data obtained from the defect fingerprint database and orthogonal test method, a mapping relationship between defect features and aging state is established using a deep learning algorithm; The high thermal conductivity epoxy resin composite material sample was tested according to the mapping relationship, and the sample evaluation results were obtained. The sample evaluation results include quantitative detection results and quantitative evaluation results; The quantitative detection results include the type, location, and size of defects in the high thermal conductivity epoxy resin composite material sample; The quantitative assessment results include the aging degree of the high thermal conductivity epoxy resin composite material sample and the predicted value of its remaining life.

[0027] Further, in step S3, the material design parameters are corrected based on the sample evaluation results to obtain a material performance dataset, and the dry-type transformer structure is designed by configuring the equivalent physical property parameters of the high thermal conductivity epoxy resin coil. Specifically: Based on the quantitative detection results in the sample evaluation results, the filler selection and interface modifier parameters in the high thermal conductivity epoxy resin formulation are optimized by feedback to obtain the corrected material design parameters. Based on the revised material design parameters and the aging degree of the high thermal conductivity epoxy resin composite material specimens in the quantitative evaluation results, a material property dataset including thermal conductivity, electrical strength, and flexural strength is established. Based on the aforementioned material property dataset, the equivalent physical property parameters of a high thermal conductivity epoxy resin coil are configured by studying the thermal conductivity characteristics under different coil winding methods and air passage arrangement methods. Based on the equivalent physical property parameters and the results of the study on the thermal conductivity characteristics of the core and coil, the overall structure of the dry-type transformer is designed, resulting in a dry-type transformer structure that includes the core structure, coil arrangement, and air duct layout.

[0028] Further, in step S3, based on the material property dataset, by studying the thermal conductivity characteristics under different coil winding methods and air passage arrangements, the equivalent physical property parameters of the high thermal conductivity epoxy resin coil are configured, specifically: Based on the thermal conductivity in the material property dataset, a porous medium free fluid coupled flow heat transfer model is established, which includes wires, insulating films, high thermal conductivity epoxy resin composite materials, and cooling air channels. Based on the porous medium free fluid coupled flow heat transfer model, and considering the gas flow characteristics under different coil winding methods (layered winding, segmented winding), and different air passage spacing and number arrangement, the gas permeability and heat dispersion coefficient are determined. Based on the gas permeability and heat dispersion coefficient, the thermal conductivity parameters of the high thermal conductivity epoxy resin coil are determined by solving the non-thermal equilibrium flow heat transfer equation and considering the coil temperature distribution characteristics under different winding and arrangement methods. Based on the aforementioned thermal conductivity parameters, and configured according to constraints including coil temperature rise limits, main insulation strength, and inter-turn insulation strength, the equivalent physical property parameters of the high thermal conductivity epoxy resin coil are obtained.

[0029] Further, in step S3, based on the dry-type transformer structure and the harmonics and power output fluctuations in the renewable energy access scenario, a multi-physics coupled simulation model is constructed and its performance is verified to obtain the performance evaluation results of the dry-type transformer. Specifically: Based on the dry-type transformer structure and the equivalent physical property parameters of the high thermal conductivity epoxy resin coil, a multi-physics field coupling simulation model including electric field, magnetic field, temperature field and stress field is constructed. Based on the harmonic frequency, harmonic content, and power output fluctuation amplitude parameters of the renewable energy access scenario, the boundary conditions and excitation sources of the multiphysics coupling simulation model are set. Based on the boundary conditions and excitation sources of the multiphysics coupling simulation model, the multiphysics coupling simulation model is run and electric field analysis is performed to obtain the electric field intensity distribution and inter-turn voltage distribution. Based on the boundary conditions and excitation source of the multiphysics coupling simulation model, the multiphysics coupling simulation model is run and magnetic field analysis is performed to obtain the magnetic flux density distribution and core loss. Based on the boundary conditions and excitation sources of the multiphysics coupled simulation model, the multiphysics coupled simulation model is run and the temperature field is analyzed to obtain the temperature distribution and hot spot temperature rise. Based on the boundary conditions and excitation sources of the multiphysics coupled simulation model, the multiphysics coupled simulation model is run and stress field analysis is performed to obtain the mechanical stress distribution and deformation displacement. Based on the electric field intensity distribution, inter-turn voltage distribution, magnetic flux density distribution, temperature distribution, hot spot temperature rise, mechanical stress distribution, and deformation displacement, the performance evaluation results of the dry-type transformer are verified.

[0030] Further, in step S3, based on the electric field intensity distribution, inter-turn voltage distribution, magnetic flux density distribution, temperature distribution, hot spot temperature rise, mechanical stress distribution, and deformation displacement, the performance evaluation results of the dry-type transformer are verified, specifically: Based on the electric field strength distribution, check whether the maximum electric field strength is lower than the breakdown field strength threshold of the high thermal conductivity epoxy resin composite material, and obtain the electric field strength check result; Based on the inter-turn voltage distribution, check whether the inter-turn potential difference meets the insulation coordination requirements, and obtain the inter-turn insulation check result; Based on the temperature distribution and hot spot temperature rise, check whether the temperature rise of the highest temperature point is lower than the preset temperature rise limit to obtain the temperature rise performance verification result. Based on the mechanical stress distribution and deformation displacement, check whether the maximum stress value is lower than the yield strength of the high thermal conductivity epoxy resin composite material and whether the deformation is within the allowable range, and obtain the mechanical stability check result; Based on the electric field strength verification results, inter-turn insulation verification results, temperature rise performance verification results, and mechanical stability verification results, a comprehensive evaluation is conducted to obtain the performance assessment results of the dry-type transformer: If all verification results pass, the dry-type transformer is deemed to meet the operating requirements, and the performance evaluation result is deemed qualified. If any verification result fails, the dry-type transformer is deemed not to meet the operating requirements, and the performance evaluation result is deemed unqualified. Based on the failed verification results, the specific substandard performance indicators and their distribution areas in the dry-type transformer structure are identified. The performance indicators include electric field strength, inter-turn insulation, hot spot temperature rise, or mechanical stress.

[0031] Further, in step S3, based on the performance evaluation results, with the optimization objectives of maximizing equipment capacity and minimizing volume, and with electric field strength, temperature rise limit, and mechanical stress as constraints, a multi-objective collaborative optimization is performed using the Monte Carlo iterative method and a global optimization algorithm to obtain the optimized global design parameters for the dry-type transformer. Specifically: If the performance evaluation result is qualified, the current dry-type transformer structural parameters will be used as the optimized global design parameters for the dry-type transformer. If the performance evaluation result is unqualified, the structural parameters of the dry-type transformer to be optimized are determined based on the specific unqualified performance indicators and their distribution areas in the dry-type transformer structure. The structural parameters of the dry-type transformer to be optimized include the epoxy resin content of the core, the coil winding method, the air passage layout parameters, the wire diameter and the wire end distance. Based on the structural parameters of the dry-type transformer to be optimized, a multi-objective optimization function is established with the optimization objectives of maximizing equipment capacity and minimizing volume, and with constraints of electric field strength not exceeding the standard, temperature rise not exceeding the limit, and mechanical stress less than the material tolerance value. Based on the multi-objective optimization function, the Monte Carlo iteration method is used to randomly sample within the feasible region of the dry-type transformer structural parameters to be optimized, generating multiple candidate dry-type transformer structural parameter design schemes. According to the global optimization algorithm, the optimization calculation is performed on the multiple candidate dry-type transformer structural parameter design schemes to obtain the candidate dry-type transformer structural parameter design schemes that satisfy all constraints and make the optimization objective optimal, which are used as the optimized global design parameters of the dry-type transformer.

[0032] Further, in step S3, a prototype is manufactured according to the optimized global design parameters of the dry-type transformer, and a power frequency-harmonic composite current temperature rise test and a high-frequency oscillation pulse voltage inter-turn insulation test are performed sequentially to obtain a high thermal conductivity epoxy resin dry-type transformer prototype that has been tested and verified. Specifically: Based on the optimized global design parameters of the dry-type transformer, a high thermal conductivity epoxy resin dry-type transformer prototype was manufactured by using high thermal conductivity epoxy resin composite material for coil winding, core assembly and vacuum casting and curing. According to the high thermal conductivity epoxy resin dry molding machine, a power frequency-harmonic composite current temperature rise test was conducted, including: A high-voltage synthetic circuit is constructed by using the harmonic compensation method, which superimposes the 3rd, 5th, and 7th characteristic harmonic currents onto the power frequency current. Pre-embedded thermocouples and infrared thermal imagers were used to measure the temperature rise distribution of the windings, and resistance strain gauges and Raman distributed fiber optic sensors were used to monitor changes in thermal stress. Run continuously until the temperature rise stabilizes, record temperature and stress data, establish the harmonic-temperature rise correspondence and thermal stress distribution model, and obtain the temperature rise and stress assessment results; A high-frequency oscillating pulse voltage inter-turn insulation test was performed using the aforementioned high thermal conductivity epoxy resin dry-type molding machine, including: An oscillating pulse voltage is generated using a high-voltage synthesis circuit with a trigger gap; The partial discharge initiation voltage was measured by boosting the voltage at a rate of 1kV / s, and lightning impulse and switching impulse voltages were applied. By monitoring the discharge signal with a partial discharge detector, the coil was disassembled after the test to observe the carbonization points, an insulation failure early warning threshold was determined, and the inter-turn insulation assessment results were obtained. Based on the temperature rise and stress assessment results and the inter-turn insulation assessment results, a comprehensive determination is made as to whether the high thermal conductivity epoxy resin dry converter prototype passes the test verification: If the test verification is passed, the high thermal conductivity epoxy resin dry deformation sampler shall be regarded as the high thermal conductivity epoxy resin dry deformation sampler verified by the test. If the test verification fails, the specific failure mode of the high thermal conductivity epoxy resin dry-type transformer prototype will be located based on the failure to meet the temperature rise and stress assessment results and / or the failure to meet the inter-turn insulation assessment results, and the global design parameters of the dry-type transformer will be optimized accordingly.

[0033] Furthermore, in step S4, a high thermal conductivity epoxy resin composite material is used as the insulating and adhesive medium to wind a coil with radial heat dissipation channels. Specifically: During the coil winding process, strip-shaped supports are placed at predetermined positions between adjacent coil layers. The strip-shaped supports are made of volatile materials. End baffles are installed at both ends of the wound coil along the axial direction. The inner ring surface of the end baffles is provided with limiting grooves that are adapted to the position and shape of the strip-shaped support. The high thermal conductivity epoxy resin composite material is used to vacuum pressure impregnate the coil that has been wound and has end baffles installed, so that the high thermal conductivity epoxy resin composite material fills the gaps between coils and turns. The coil after vacuum pressure impregnation is subjected to a first-stage curing treatment. The temperature of the first-stage curing treatment is controlled within the range that allows the high thermal conductivity epoxy resin composite material to initially gel but is below the sublimation point of the volatile material. The coil, after the first stage of curing, is subjected to a second stage of gradient temperature curing. The temperature curve of the second stage of gradient temperature curing is set to allow the volatile material to completely sublimate and escape, forming a continuous radial heat dissipation channel between the coil layers, while simultaneously allowing the high thermal conductivity epoxy resin composite material to be completely cured and molded, forming a coil with radial heat dissipation channels.

[0034] Furthermore, in step S4, a high thermal conductivity epoxy resin composite material is used to bond and cure the stacked iron cores to form an integral iron core. Specifically: Multiple stamped iron chips are stacked according to a preset stacking diagram to form a stack body, so that the micro protrusions of adjacent iron chips are staggered and arranged to form a connected resin microchannel mesh inside the stack body. The iron chips have multiple micro protrusions protruding in the stacking direction at preset positions on their edges. In a vacuum environment, a high thermal conductivity epoxy resin composite material preheated to a preset viscosity range is injected and impregnated into the stack, so that the high thermal conductivity epoxy resin composite material fills the gaps between all iron chips through the resin microchannel mesh. Axial pressure is applied to the impregnated stack and gradient temperature is increased for curing. During the gradient temperature increase curing process, the thickness of the insulating layer between the iron chips is controlled and maintained uniformly through the mutual support of the micro bosses. After gradient heating and curing, the integral iron core is obtained through cooling and post-treatment.

[0035] Furthermore, the process of applying axial pressure to the impregnated stack and performing gradient temperature curing, wherein the mutual support of the micro-boobs controls and maintains the uniformity of the insulating layer thickness between the iron chips during the gradient temperature curing process, includes: In a vacuum environment, a preset first axial pressure is applied to the impregnated stack along its stacking direction and maintained for a first preset time, so that excess high thermal conductivity epoxy resin composite material is discharged from the resin microchannel mesh at the edge of the stack. After debinding, the stacked body is transferred to a curing mold, a second axial pressure higher than the first axial pressure is applied, and the stacked body is heated to a first preset temperature at a first heating rate, so that the high thermal conductivity epoxy resin composite material reaches a predetermined viscosity and begins to gel. The temperature is raised to a second preset temperature at a second heating rate and held for a second preset time. The change in the thickness of the stack is monitored in real time by the displacement sensor built into the curing mold, and the second axial pressure is dynamically adjusted based on the change to compensate for the shrinkage of the high thermal conductivity epoxy resin composite material during the curing process and maintain the uniform thickness of the insulation layer between the iron chips.

[0036] Further, in step S4, the coil is fitted onto the integral iron core to form an assembly, and an insulating component is installed on the assembly to form an axial heat dissipation channel. Temperature and stress sensors are pre-embedded inside the coil. Specifically: After covering the surface of the core column of the integral iron core with an insulating film, elastic insulating support strips with self-centering function are installed. The elastic insulating support strips are distributed along the core column axis of the integral iron core and arranged in a circumferential array. After the coil is heated to the preset expansion temperature, it is fitted onto the core post of the integral iron core on which the elastic insulating support bar has been installed. After the coil cools and shrinks, the inner wall of the coil is interference-fitted with the elastic insulating support bar and pressed tightly to form the first positioning. Insulating end rings are fitted at both ends of the coil along the axial direction, and U-shaped groove insulating partitions made of high thermal conductivity epoxy resin composite material are inserted between adjacent coils and between the coil and the overall iron core yoke. Each U-shaped groove insulating partition and the elastic insulating support strip are interleaved to form an axial heat dissipation channel extending along the coil axis. Before mounting the coil, the temperature and stress sensors are attached to the preset hot spots and stress concentration areas on the inner wall of the coil. The leads of the temperature and stress sensors are led out along the radial heat dissipation channel and temporarily fixed at the end of the coil through a detachable sealing joint.

[0037] Furthermore, in step S4, the assembly is integrally cast using a high thermal conductivity epoxy resin composite material in a vacuum environment, and then cured under gradient temperature control to produce a transformer prototype. Specifically: Before casting, the assembly is vacuum dried and the leads of the temperature and stress sensors are led out through the sealed terminals pre-set on the casting mold. The high thermal conductivity epoxy resin composite material, preheated to the casting viscosity, is injected into the casting mold in two stages under vacuum: In the first stage, the high thermal conductivity epoxy resin composite material is injected from the bottom of the casting mold at the first injection rate, so that the high thermal conductivity epoxy resin composite material preferentially fills the gaps between the coil, the integral iron core and the insulating parts at the bottom of the assembly. In the second stage, the high thermal conductivity epoxy resin composite material is injected from the top of the casting mold at a second injection rate higher than the first injection rate until the assembly is completely submerged in the high thermal conductivity epoxy resin composite material. After the casting is completed, the first negative pressure value is maintained in a vacuum environment and the mixture is left to stand for a first preset time to allow the air bubbles in the high thermal conductivity epoxy resin composite material to escape. Gradient temperature-controlled curing is performed on the casting mold and its internal materials after casting. The temperature of the casting mold and its internal material is raised to the first curing temperature at a first heating rate and kept at that temperature to allow the high thermal conductivity epoxy resin composite material to initially gel. The temperature of the casting mold and the internal material is raised to the second curing temperature at a second heating rate lower than the first heating rate and kept at the temperature to allow the high thermal conductivity epoxy resin composite material to fully cure. The temperature of the casting mold and the internal material is then slowly cooled to the demolding temperature at a controlled cooling rate. During the heat preservation stage of gradient temperature-controlled curing, the heat preservation temperature or heat preservation time is adjusted based on the real-time monitoring of temperature and stress changes of the internal material using temperature and stress sensors led out from the sealed terminal.

[0038] Furthermore, the transformer prototype testing method described in step S4 includes: An oscillating pulse voltage wave simulating frequent overvoltage conditions is applied to the high-voltage winding of the transformer prototype, while a power frequency voltage is applied to the low-voltage winding. The inter-turn insulation withstand data of the transformer prototype are continuously acquired and recorded at a preset period. A composite current containing the fundamental wave and a specified harmonic is applied to a transformer prototype to simulate load fluctuations caused by renewable energy access. Temperature rise distribution, hot spot data and mechanical vibration spectrum of the transformer prototype are collected and generated by pre-embedded temperature and stress sensors. Based on the inter-turn insulation tolerance data, temperature rise distribution, hot spot data, and mechanical vibration spectrum, the thermal conductivity, insulation performance, and mechanical performance of the transformer prototype are evaluated. The transformer prototype was manufactured using the aforementioned method.

[0039] Furthermore, the step of applying oscillating pulse voltage waves simulating frequent overvoltage conditions to the high-voltage winding of the transformer prototype, while simultaneously applying power frequency voltage to the low-voltage winding, and continuously acquiring and recording the inter-turn insulation withstand data of the transformer prototype for a preset period, includes: The transformer prototype was placed in a shielded test environment, and the first end of the high voltage winding was connected to the output end of the high voltage pulse generator. The first and last ends of the low voltage winding were short-circuited and connected to one pole of the power frequency voltage source. At the same time, the core and shell of the transformer prototype were grounded. An oscillating pulse voltage wave is applied to the high-voltage winding at a preset repetition frequency and pulse amplitude. The waveform of the oscillating pulse voltage wave is a decaying oscillating waveform, the voltage value of its first peak is a preset multiple of the peak value of the rated phase voltage of the transformer prototype, and its oscillation frequency is within a preset range of the inherent resonant frequency of the transformer prototype winding. While applying an oscillating pulse voltage wave to the high-voltage winding, a power frequency sinusoidal voltage equal to the rated voltage of the transformer prototype is applied to the low-voltage winding, and the phase of the power frequency sinusoidal voltage is controlled so that its peak voltage moment is synchronized with the first peak moment of the oscillating pulse voltage wave. A complete test sequence is formed by continuously applying voltage combinations a preset number of times. In each test sequence, the pulse impact current waveform and partial discharge signal of the high-voltage winding, as well as the power frequency leakage current of the low-voltage winding, are collected synchronously. The transfer function of the winding is calculated based on the pulse impact current waveform, and the inter-turn insulation withstand data of the transformer prototype is analyzed and recorded according to the changes in the partial discharge signal and the power frequency leakage current.

[0040] Furthermore, the application of a composite current containing the fundamental wave and a specified harmonic to the transformer prototype to simulate load fluctuations from renewable energy integration, and the acquisition and generation of temperature rise distribution, hotspot data, and mechanical vibration spectrum of the transformer prototype through pre-embedded temperature and stress sensors, includes: A test current, composed of a power frequency fundamental wave and a specified harmonic with a preset amplitude and phase, is applied to the low-voltage winding of the transformer prototype using a programmable composite current source to simulate a typical harmonic spectrum with a power electronic converter connected. The programmable composite current source is controlled to output the test current according to a preset load cycle curve. In the load cycle curve, the amplitude of the fundamental current changes periodically, and the content of each harmonic is adjusted according to a preset ratio to simulate the load fluctuation of renewable energy access. During the application of the test current, data from temperature and stress sensors embedded inside the coil are collected simultaneously, and vibration acceleration signals are collected at preset monitoring points on the outer shell of the transformer prototype. The collected temperature data is mapped and fitted to the three-dimensional structural model of the transformer prototype to generate a dynamic temperature rise distribution map during the operation of the transformer prototype, and the temperature area exceeding the preset threshold is identified as hot spot data. Time-frequency analysis was performed on the collected stress data and vibration acceleration signals to extract characteristic frequency components related to the specified harmonic frequency, and the mechanical vibration spectrum of the transformer prototype under harmonic excitation was synthesized.

[0041] Furthermore, the evaluation of the thermal conductivity, insulation performance, and mechanical performance of the transformer prototype based on inter-turn insulation tolerance data, temperature rise distribution, hot spot data, and mechanical vibration spectrum includes: Based on the dynamic temperature rise distribution map and the hot spot data, the thermal conductivity of the transformer prototype is evaluated using computer-equivalent thermal network parameters. Based on the inter-turn insulation tolerance data, a quantitative index of insulation aging degree is calculated to evaluate the insulation performance of the transformer prototype. The vibration spectrum of the structure is compared with the theoretical spectrum obtained by modal simulation based on the three-dimensional model of the transformer prototype to evaluate the mechanical performance of the transformer prototype.

[0042] Secondly, this application also provides a comprehensive preparation and performance assurance system for a high thermal conductivity epoxy resin-based dry-type transformer, applied to the aforementioned comprehensive preparation and performance assurance method for the high thermal conductivity epoxy resin-based dry-type transformer, comprising: The performance control module 100 is used to control the performance of the high thermal conductivity epoxy resin composite material according to the preset material performance target value by using a performance control method based on the high thermal conductivity epoxy resin composite material to obtain optimized material design parameters. The material design parameters include epoxy resin matrix parameters, compound thermally conductive filler parameters, and interface modifier parameters. The detection and evaluation module 200 is used to detect and evaluate a high thermal conductivity epoxy resin composite material sample prepared according to the material design parameters using a detection and evaluation method based on the high thermal conductivity epoxy resin composite material, obtain the sample evaluation result, and correct the material design parameters according to the sample evaluation result to obtain the corrected material design parameters. The design and verification module 300 is used to adopt the high thermal conductivity epoxy resin dry transformer design and verification method, and perform multi-scale modeling, structural design, multi-physics field coupling simulation verification and multi-objective collaborative optimization of the dry transformer according to the modified material design parameters, so as to obtain the optimized global design parameters of the dry transformer. The prototype manufacturing module 400 is used to manufacture a high thermal conductivity epoxy resin-based dry-type transformer prototype according to the optimized global design parameters of the dry-type transformer using the transformer prototype manufacturing method, and to conduct comprehensive performance tests and verifications on the transformer prototype using the transformer prototype testing method, thereby completing the comprehensive preparation and performance assurance of the high thermal conductivity epoxy resin-based dry-type transformer. The performance control method is used to determine the filler combination, interface modifier, and related interface and structural parameters of the high thermal conductivity epoxy resin composite material, so as to achieve precise control of the macroscopic thermal conductivity, electrical strength, and flexural strength of the material; the detection and evaluation method is used to quantitatively detect the defect type, location, and size of the composite material sample, and evaluate the aging degree and remaining life of the sample; the dry-type transformer design and verification method is used to complete the microscopic, mesoscopic, and macroscopic multi-scale modeling of the dry-type transformer, as well as the structural design and multi-physics field coupling simulation verification of the dry-type transformer based on the renewable energy access scenario; the transformer prototype manufacturing method is used to complete the molding and overall casting and curing of the transformer coil and core; and the transformer prototype testing method is used to verify the inter-turn insulation, thermal conductivity, and mechanical properties of the transformer prototype.

[0043] Thirdly, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method for the comprehensive preparation and performance assurance of a high thermal conductivity epoxy resin-based dry-type transformer.

[0044] Compared with the prior art, this application has the following beneficial effects: This application optimizes high thermal conductivity epoxy resin composite materials through a cross-scale collaborative control mechanism of "macro-meso-micro". It constructs a multi-physics field coupling simulation model of "electro-magnetic-thermal-mechanical" for collaborative design, and fabricates a dry-type transformer prototype with radial-axial dual heat dissipation channels and embedded sensors. It uses broadband electromagnetic wave scanning technology to accurately detect microcrack defects and assess aging, and conducts a series of performance tests simulating complex operating conditions of renewable energy. Finally, by combining the detection and assessment results with the test data, and through closed-loop verification and iterative optimization, it achieves precise control of the dry-type transformer's performance throughout its entire life cycle. Attached Figure Description

[0045] Figure 1 A schematic flowchart illustrating the comprehensive preparation and performance assurance method of a high thermal conductivity epoxy resin-based dry-type transformer provided in the embodiments of this application; Figure 2 This is a schematic diagram of the integrated fabrication and performance assurance system for a high thermal conductivity epoxy resin-based dry-type transformer provided in the embodiments of this application. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as being processed sequentially, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. A process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. A process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0047] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0048] Existing dry-type transformers face complex operating conditions involving the coupling of electric, magnetic, temperature, and stress fields when renewable energy is connected to the grid on a large scale. These problems include difficulty in accurately matching the thermal conductivity, insulation, and mechanical properties of materials to preset targets; the inability of single-physics field verification modes to simulate multi-field coupling effects; simplistic heat dissipation design; insufficient structural integration; manufacturing processes that easily lead to structural inhomogeneity; residual bubbles; insufficient accuracy in detecting microcracks; disconnect between aging tests and actual operating conditions; and the lack of a closed loop of "design-manufacturing-testing-optimization." Consequently, they are unable to meet the reliability verification requirements of complex operating scenarios.

[0049] In this regard, such as Figure 1 As shown in the figure, this embodiment proposes a comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer, including the following steps: S1. A performance control method based on high thermal conductivity epoxy resin composite material is adopted. The performance of high thermal conductivity epoxy resin composite material is controlled according to the preset material performance target value to obtain optimized material design parameters. The material design parameters include epoxy resin matrix parameters, compound thermally conductive filler parameters, and interface modifier parameters. S2. Using a detection and evaluation method based on high thermal conductivity epoxy resin composite materials, the high thermal conductivity epoxy resin composite material samples prepared according to the material design parameters are detected and evaluated to obtain the sample evaluation results. Based on the sample evaluation results, the material design parameters are corrected to obtain the corrected material design parameters. S3. Using the high thermal conductivity epoxy resin dry transformer design and verification method, based on the modified material design parameters, the dry transformer undergoes multi-scale modeling, structural design, multi-physics field coupling simulation verification, and multi-objective collaborative optimization to obtain the optimized global design parameters of the dry transformer. S4. Using the transformer prototype manufacturing method, a high thermal conductivity epoxy resin-based dry-type transformer prototype is manufactured according to the optimized global design parameters of the dry-type transformer. The transformer prototype is then tested and verified using the transformer prototype testing method to complete the comprehensive preparation and performance assurance of the high thermal conductivity epoxy resin-based dry-type transformer. The performance control method is used to determine the filler combination, interface modifier, and related interface and structural parameters of the high thermal conductivity epoxy resin composite material, so as to achieve precise control of the macroscopic thermal conductivity, electrical strength, and flexural strength of the material; the detection and evaluation method is used to quantitatively detect the defect type, location, and size of the composite material sample, and evaluate the aging degree and remaining life of the sample; the dry-type transformer design and verification method is used to complete the microscopic, mesoscopic, and macroscopic multi-scale modeling of the dry-type transformer, as well as the structural design and multi-physics field coupling simulation verification of the dry-type transformer based on the renewable energy access scenario; the transformer prototype manufacturing method is used to complete the molding and overall casting and curing of the transformer coil and core; and the transformer prototype testing method is used to verify the inter-turn insulation, thermal conductivity, and mechanical properties of the transformer prototype.

[0050] Further, in step S1, based on the preset performance target value, a model parameter set is obtained by screening filler combination schemes and corresponding interface modifiers through high-throughput calculations. Specifically: Based on a pre-set filler database, a performance prediction proxy model is trained using a machine learning algorithm. The inputs of the performance prediction proxy model include filler type, filler shape, filler volume fraction, and molecular descriptor of the interface modifier. Using the performance target value as the optimization objective, a multi-objective optimization algorithm is used to perform a global search in the prediction space of the performance prediction proxy model to select the Pareto optimal solution set. By extracting common features from the Pareto optimal solution set, the filler combination scheme and the corresponding interface modifier are determined, forming a model parameter set.

[0051] Furthermore, the step of determining the filler combination scheme and corresponding interface modifier by extracting common features from the Pareto optimal solution set to form a model parameter set includes: The Pareto optimal solution set is grouped using a clustering algorithm. The characteristic variables of the clustering algorithm include filler type, filler shape, filler gradation ratio, and molecular descriptor of interface modifier. Based on each characteristic variable, a characteristic statistical matrix is ​​established to quantitatively calculate the average aspect ratio of the filler, the variance of the aspect ratio distribution, the gradation ratio of the filler, and the numerical concentration interval of the molecular descriptor of the interface modifier. Based on the feature statistics matrix, principal component analysis is used to identify the feature combinations that affect the performance of high thermal conductivity epoxy resin composites, and filler combination schemes that appear repeatedly in multiple Pareto optimal solutions and whose frequency exceeds a preset threshold are selected. The structural parameters of the interface modifier most strongly correlated with the filler combination scheme were determined by correlation analysis. The structural parameters include molecular chain length, type and number of polar groups, and molecular conformation characteristics. A model parameter set is constructed based on the selected filler combination schemes and the structural parameters of their corresponding interface modifiers.

[0052] Furthermore, in step S1, the optimal interface parameters are obtained by calculating the interface binding energy and interface thermal conductivity through molecular dynamics simulation based on the model parameter set. Specifically: Based on the structural parameters of each filler combination scheme and its corresponding interface modifier in the model parameter set, a three-component atomic model including epoxy resin matrix, filler surface and interface modifier is constructed. Energy minimization and molecular dynamics relaxation simulations were performed sequentially on each three-component atomic model to obtain a stable equilibrium system structure. Based on the aforementioned equilibrium system structure, the interfacial binding energy corresponding to each filler combination scheme is calculated using energy analysis methods in molecular dynamics simulations. The interfacial binding energy is used to characterize the adsorption strength between the interfacial modifier and the filler surface. Based on the non-equilibrium molecular dynamics method, a temperature gradient is established at both ends of the equilibrium system structure, and the interfacial thermal conductivity corresponding to each packing combination scheme is calculated by the steady-state heat flow method. Based on the calculation results of interfacial binding energy and interfacial thermal conductivity corresponding to all filler combination schemes, quantitative correlation models between the molecular structure parameters of the interfacial modifier and the interfacial binding energy and interfacial thermal conductivity were established by multiple regression analysis. Based on the quantitative correlation model, a multi-objective optimization algorithm is used to iteratively optimize the molecular structure parameters of the interface modifier with the goal of simultaneously maximizing the interfacial binding energy and interfacial thermal conductivity, so as to obtain the optimal interfacial parameters.

[0053] Furthermore, based on the calculated interfacial binding energy and interfacial thermal conductivity of all filler combination schemes, quantitative correlation models between the molecular structure parameters of the interfacial modifier and the interfacial binding energy and interfacial thermal conductivity are established through multiple regression analysis, including: The molecular structure parameters of the interface modifier are used as the set of independent variables, and the interface binding energy and interface thermal conductivity are used as the dependent variables, respectively, to construct an initial multivariate regression dataset. The initial multivariate regression dataset is standardized, and the variance inflation factor method is used to diagnose multicollinearity. Independent variables with variance inflation factors greater than a preset threshold are removed to obtain the processed dataset. Based on the processed dataset, the stepwise regression method was used to screen the molecular structure parameters of the interface modifiers with the p-value of the statistical test being less than a preset threshold, and an initial quantitative correlation model was established. The prediction accuracy of the initial quantitative correlation model was evaluated using k-fold cross-validation. Based on the evaluation results, the coefficients of the initial quantitative correlation model were regularized and optimized using ridge regression algorithm to obtain a quantitative correlation model between the molecular structure parameters of the interface modifier and the interface binding energy and interface thermal conductivity.

[0054] Furthermore, in step S1, based on the optimal interface parameters, the mesoscopic structural characteristic parameters are obtained by simulating the packing dispersion process and interface layer evolution process using the phase-field method. Specifically: Based on the optimal interface parameters, a three-phase phase field model is established, which includes an epoxy resin matrix phase, a filler phase, and an interface modifier phase. Three sets of sequence parameters are defined to characterize the distribution state of the epoxy resin matrix phase, the filler phase, and the interface modifier phase, respectively. Based on the interface bonding energy and interface thermal conductivity in the optimal interface parameters, the interface energy parameters and gradient energy coefficients between each phase in the three-phase phase field model are set. Run the three-phase phase field model until the system reaches a dynamic equilibrium state; The spatial distribution, orientation distribution, packing network shape, and interface layer thickness distribution of the packing under dynamic equilibrium conditions are extracted as simulation results. Based on the simulation results, the permeation threshold, thermal conductivity path density, average thickness of the interface layer and its distribution uniformity of the packing network are quantitatively calculated to form a set of mesoscopic structural feature parameters that include packing distribution characteristics and interface layer characteristics.

[0055] Furthermore, running the three-phase phase-field model until the system reaches a dynamic equilibrium state includes: In the three-phase phase field model, a convection term describing the shear flow field is introduced. By coupling the flow field control equation and the phase field evolution equation, the dispersion process, migration process and agglomeration kinetics of the filler in the epoxy resin matrix are simulated until the system reaches a dynamic equilibrium state. Based on the optimal interface parameters, the interface gradient parameters are set, and the adsorption process of the interface modifier on the filler surface and the formation and evolution process of the interface layer are simulated by solving the interface evolution equation until the system reaches a dynamic equilibrium state.

[0056] Furthermore, the convection term describing the shear flow field is introduced into the three-phase phase field model. By coupling the flow field control equation and the phase field evolution equation, the dispersion, migration, and aggregation dynamics of the filler in the epoxy resin matrix are simulated until the system reaches a dynamic equilibrium state, including: Based on the actual processing parameters, the flow field parameters in the three-phase phase field model are initialized, including the shear rate, viscosity and density of the epoxy resin matrix; Establish the coupled flow field control equations and phase field evolution equations; A convection term describing the shear flow field is introduced into the phase field evolution equation. The convection term is composed of the product of the flow field velocity vector and the order parameter gradient, and is used to describe the influence of the flow field on the packing transport and interface evolution. The coupled equations are spatially discretized using the finite difference method, and time integration is performed using an implicit-explicit hybrid time-progression method. Within each time step, the flow field control equations are first solved to obtain the updated flow field velocity distribution. Then, the flow field velocity distribution is substituted as a known quantity into the phase field evolution equations to solve for the order parameters of each time step. Based on the sequence parameters at each time step, the dispersion, migration and aggregation kinetics of fillers in the epoxy resin matrix are simulated by an iterative method until the system reaches a dynamic equilibrium state. The process of setting interface gradient parameters based on the optimal interface parameters, simulating the adsorption process of the interface modifier on the filler surface and the formation and evolution of the interface layer by solving the interface evolution equation, until the system reaches a dynamic equilibrium state, includes: Based on the interfacial binding energy in the optimal interfacial parameters, the adsorption energy parameters of the interfacial modifier on the filler surface are determined, and the interfacial gradient energy coefficient in the three-phase phase field model is set based on the adsorption energy parameters. Initialize the concentration distribution field of the interface modifier in the epoxy resin matrix and set adsorption boundary conditions on the filler surface; A non-conservative Allen-Cahn equation describing the evolution of the interface layer is established. The non-conservative Allen-Cahn equation includes an interface gradient term, a double-well potential function term, and a convection transport term. Based on the interfacial gradient energy coefficient and the adsorption energy parameter, the interfacial evolution equation is solved to simulate the adsorption process of the interfacial modifier on the filler surface and the formation and evolution process of the interfacial layer until the system reaches a dynamic equilibrium state.

[0057] Further, in step S1, the macroscopic thermal conductivity, electrical strength, and flexural strength of the high thermal conductivity epoxy resin composite material are predicted by finite element analysis based on the mesoscopic structural characteristic parameters, resulting in predicted performance values. Specifically: Based on the filler spatial distribution, orientation distribution, and filler network shape in the mesoscopic structural characteristic parameters, a three-dimensional geometric model of the high thermal conductivity epoxy resin composite material is reconstructed. Based on the filler type and the thickness distribution of the interface layer, the filler phase, the interface layer and the epoxy resin matrix phase are respectively assigned corresponding material properties, including thermal conductivity, dielectric constant, electrical conductivity and elastic modulus; The three-dimensional geometric model is meshed, and finite element models of the heat conduction control equation, the electrostatic conduction control equation, and the linear elasticity control equation are established respectively. Temperature boundary conditions are applied to the finite element model of heat conduction, and the temperature field distribution is obtained by solving. The macroscopic thermal conductivity of the high thermal conductivity epoxy resin composite material is calculated based on Fourier's law. Voltage boundary conditions are applied to the finite element model of electrostatic conduction, and the electric field distribution is obtained by solving. The electrical strength of the high thermal conductivity epoxy resin composite material is predicted based on the extreme values ​​of the electric field strength. Displacement and stress boundary conditions are applied in the finite element model of online elasticity, and the stress-strain field is obtained by solving. The bending strength of high thermal conductivity epoxy resin composite material is predicted based on the maximum stress criterion. The macroscopic thermal conductivity, electrical strength, and flexural strength are used as performance prediction values.

[0058] Further, in step S1, the performance is controlled by adjusting the filler volume fraction and interface modification strength based on the difference between the predicted performance value and the target performance value. Specifically: Calculate the relative error between the predicted performance value and the target performance value, wherein the predicted performance value includes the predicted macroscopic thermal conductivity, the predicted electrical strength, and the predicted flexural strength; A multi-objective optimization function is established based on the relative error, with the optimization objective being to minimize the weighted sum of the relative errors of the predicted macroscopic thermal conductivity, predicted electrical strength, and predicted flexural strength. Key control parameters for macroscopic thermal conductivity, electrical strength and flexural strength are identified. These key control parameters include filler volume fraction and interface modification strength, wherein the interface modification strength is characterized by the functional group density and molecular chain length of the interface modifier. Based on gradient descent or genetic algorithm, the filler volume fraction and interface modification intensity are iteratively optimized within a preset parameter adjustment range, wherein the interface modification intensity is characterized by the functional group density and molecular chain length of the interface modifier. When the value of the multi-objective optimization function converges within the preset tolerance range, or reaches the maximum number of iterations, the optimization process is stopped, and the optimal combination of filler volume fraction and interface modification strength is obtained. Performance is controlled based on the optimal combination of filler volume fraction and interfacial modification strength.

[0059] For example: This embodiment introduces a method for performance control based on high thermal conductivity epoxy resin composite materials, including: Step 1: Based on the preset performance target values, obtain the model parameter set by screening the filler combination scheme and the corresponding interface modifier through high-throughput calculation.

[0060] This embodiment uses high-throughput computation to screen filler combination schemes and interface modifiers, which can quickly construct a model parameter set that meets the preset performance target values, significantly improve the efficiency of high thermal conductivity epoxy resin composite material formulation screening, reduce experimental trial and error costs, and realize rapid mapping from multi-dimensional parameters to precise formulations.

[0061] Step 2: Based on the model parameter set, calculate the interface binding energy and interface thermal conductivity through molecular dynamics simulation to obtain the optimal interface parameters.

[0062] This embodiment uses molecular dynamics simulation based on model parameter set, which can accurately quantify the intrinsic relationship between interfacial binding energy and interfacial thermal conductivity, optimize interfacial parameter design, provide theoretical guidance for the selection of interfacial modifiers and interfacial structure design, and improve the accuracy of interfacial thermal conductivity performance control.

[0063] Step 3: Based on the optimal interface parameters, the mesoscopic structural characteristic parameters are obtained by simulating the dispersion process of the filler and the evolution process of the interface layer using the phase field method.

[0064] This embodiment uses the phase field method to simulate the filler dispersion process and interface layer evolution, which can dynamically capture the filler distribution law and interface structure evolution characteristics at the mesoscale, obtain mesoscale structural characteristic parameters, and provide key structural basis for the correlation analysis of the microstructure and macroscopic properties of high thermal conductivity epoxy resin composites.

[0065] Step 4: Based on the mesoscopic structural characteristic parameters, predict the macroscopic thermal conductivity, electrical strength, and flexural strength of the high thermal conductivity epoxy resin composite material through finite element analysis to obtain the predicted performance values.

[0066] This embodiment combines mesoscopic structural characteristic parameters with finite element analysis to establish a predictive model from mesoscopic structure to macroscopic performance, accurately predicting the thermal conductivity, electrical strength and flexural strength of high thermal conductivity epoxy resin composites, and realizing digital prediction and verification of performance.

[0067] Step 5: Adjust the filler volume fraction and interface modification strength to regulate performance based on the difference between the predicted performance value and the target performance value.

[0068] This embodiment dynamically adjusts the filler volume fraction and interface modification strength based on the difference between the predicted performance value and the target value, which can form a closed-loop feedback mechanism for performance regulation and ensure that the final high thermal conductivity epoxy resin composite material performance accurately meets the preset target value.

[0069] The implementation steps of a performance control method based on high thermal conductivity epoxy resin composite materials include: Step 1: Based on the preset performance target values, obtain the model parameter set by screening the filler combination scheme and the corresponding interface modifier through high-throughput calculation.

[0070] Step 1.1: Based on the pre-set filler database, a performance prediction proxy model is trained using machine learning algorithms.

[0071] In this embodiment, the inputs to the performance prediction proxy model include filler type, filler shape, filler volume fraction, and molecular descriptor of the interface modifier.

[0072] This embodiment constructs a performance prediction proxy model based on a filler database and machine learning algorithms. It can realize rapid correlation mapping of multi-dimensional parameters such as filler type, shape, volume fraction and interface modifier molecular descriptor, significantly improve the efficiency of material performance prediction, provide accurate prediction support for subsequent multi-objective optimization, and reduce the cost and cycle of traditional experimental trial and error.

[0073] Step 1.2: Using the performance target value as the optimization objective, a multi-objective optimization algorithm is used to perform a global search in the prediction space of the performance prediction proxy model to select the Pareto optimal solution set.

[0074] This embodiment uses the performance target value as the optimization objective. Through a multi-objective optimization algorithm, a global search is performed in the prediction model space. This can efficiently screen out the non-dominated Pareto optimal solution set, balance the competitive relationship of multiple performance indicators such as thermal conductivity, electrical strength, and flexural strength, and provide a basis for multiple scheme optimization for the selection of filler combination and interface modifier.

[0075] Step 1.3: Determine the filler combination scheme and the corresponding interface modifier by extracting the common features from the Pareto optimal solution set, and form a model parameter set.

[0076] This embodiment determines the filler combination and interface modifier by extracting the common features of the Pareto optimal solution set, which can form a structured model parameter set, realize the key transition from the optimized solution set to the engineering formulation, and improve the engineering applicability and performance reproducibility of the parameter set.

[0077] Step 1.3.1: The Pareto optimal solution set is grouped using a clustering algorithm. The characteristic variables of the clustering algorithm include filler type, filler shape, filler gradation ratio, and molecular descriptor of interface modifier.

[0078] This embodiment uses a clustering algorithm to group the Pareto optimal solution set. It can achieve dynamic classification based on characteristic variables such as filler type, filler shape, filler gradation ratio, and molecular descriptor of interface modifier, revealing the inherent distribution law of different filler combination schemes.

[0079] Step 1.3.2: Establish a feature statistical matrix based on each feature variable, and quantitatively calculate the average aspect ratio of the filler, the aspect ratio distribution variance, the filler gradation ratio, and the numerical concentration interval of the molecular descriptor of the interface modifier.

[0080] This embodiment quantifies the average aspect ratio, aspect ratio distribution variance, packing gradation ratio, and numerical concentration range of molecular descriptors of interface modifiers by establishing a characteristic statistical matrix. This can accurately quantify the statistical characteristics of mesoscopic structural parameters and provide data support for the quantitative analysis of packing dispersibility and interfacial binding energy.

[0081] Step 1.3.3: Based on the feature statistics matrix, principal component analysis is used to identify the feature combinations that affect the performance of high thermal conductivity epoxy resin composites, and filler combination schemes that appear repeatedly in multiple Pareto optimal solutions and whose frequency exceeds a preset threshold are selected.

[0082] This embodiment uses principal component analysis based on feature statistics matrix to identify key feature combinations, which can screen out filler combination schemes that frequently appear in multiple Pareto optimal solutions, uncover the core factors affecting the performance of high thermal conductivity epoxy resin composite materials, and improve the representativeness and prediction accuracy of the model parameter set.

[0083] Step 1.3.4: Determine the structural parameters of the interface modifier that is most strongly correlated with the filler combination scheme through correlation analysis. The structural parameters include molecular chain length, type and number of polar groups, and molecular conformation characteristics.

[0084] This embodiment uses correlation analysis to determine the structural parameters of the interface modifier that are strongly correlated with the filler combination scheme, such as molecular chain length, type and number of polar groups, and molecular conformation characteristics. This can establish a quantitative correlation between filler and interface modifier, providing a theoretical basis for the accurate selection of interface modifier and optimizing the matching of interface thermal conductivity and electrical properties.

[0085] Step 1.3.5: Construct a model parameter set based on the selected filler combination schemes and the structural parameters of their corresponding interface modifiers.

[0086] This embodiment constructs a model parameter set based on the selected filler combination scheme and its corresponding interface modifier structural parameters, which can form a complete material design parameter system. This provides standardized input data for subsequent molecular dynamics simulation, phase field method simulation and finite element analysis, ensuring the consistency and traceability of performance control throughout the entire process.

[0087] Step 2: Based on the model parameter set, calculate the interface binding energy and interface thermal conductivity through molecular dynamics simulation to obtain the optimal interface parameters.

[0088] Step 2.1: Based on the structural parameters of each filler combination scheme and its corresponding interface modifier in the model parameter set, construct a three-component atomic model including epoxy resin matrix, filler surface and interface modifier.

[0089] This embodiment constructs a three-component atomic model comprising an epoxy resin matrix, a filler surface, and an interface modifier, which can accurately characterize the atomic-level interaction details of the material's microstructure, providing high-fidelity structural input for molecular dynamics simulations.

[0090] Step 2.2: Perform energy minimization and molecular dynamics relaxation simulations sequentially on each three-component atomic model to obtain a stable equilibrium system structure.

[0091] This embodiment obtains a stable equilibrium system structure through energy minimization and molecular dynamics relaxation simulation, which can eliminate local stress distortion of the initial conformation, ensure the physical authenticity of the interface binding energy and thermal conductivity calculations, and improve the reliability and repeatability of the simulation results.

[0092] Step 2.3: Based on the equilibrium system structure, the interfacial binding energy corresponding to each filler combination scheme is calculated using the energy analysis method in molecular dynamics simulation. The interfacial binding energy is used to characterize the adsorption strength between the interfacial modifier and the filler surface.

[0093] This embodiment uses energy analysis to calculate the interfacial binding energy, which can quantify the adsorption strength of the interfacial modifier and the filler surface, directly reflecting the interfacial modification effect and providing key performance indicators for the screening and optimization of interfacial modifiers.

[0094] Step 2.4: Based on the non-equilibrium molecular dynamics method, establish a temperature gradient at both ends of the equilibrium system structure, and calculate the interfacial thermal conductivity corresponding to each packing combination scheme by means of the steady-state heat flow method.

[0095] This embodiment establishes the temperature gradient and calculates the interfacial thermal conductivity based on the non-equilibrium molecular dynamics method, which can accurately quantify the heat transfer efficiency at the interface, reveal the intrinsic law of the synergistic effect between interfacial thermal conductivity and fillers and modifiers, and guide the directional optimization of interfacial thermal conductivity.

[0096] Step 2.5: Based on the calculation results of interfacial binding energy and interfacial thermal conductivity corresponding to all filler combination schemes, quantitative correlation models between the molecular structure parameters of the interfacial modifier and the interfacial binding energy and interfacial thermal conductivity are established by multiple regression analysis.

[0097] This embodiment establishes a quantitative correlation model between molecular structure parameters and interfacial properties through multiple regression analysis, which can reveal the influence weight of molecular structure parameters of interfacial modifiers on interfacial binding energy and thermal conductivity, providing a theoretical basis for the molecular design of interfacial modifiers.

[0098] Step 2.5.1: Construct an initial multivariate regression dataset, which can systematically integrate molecular structure parameters and interface performance data to form a structured predictive model input, ensuring the data integrity of model training.

[0099] This embodiment constructs an initial multivariate regression dataset, which can systematically integrate molecular structure parameters and interface performance data to form a structured predictive model input, ensuring the data integrity of model training.

[0100] Step 2.5.2: Construct an initial multiple regression dataset by using the molecular structure parameters of the interface modifier as the set of independent variables and the interface binding energy and interface thermal conductivity as the dependent variables.

[0101] This embodiment constructs an initial multiple regression dataset with the molecular structure parameters of the interface modifier as the independent variable set and the interface binding energy and interface thermal conductivity as the dependent variables. This dataset can systematically integrate the correlation data between microstructural parameters and macroscopic interface performance, forming a structured prediction model input framework. This ensures the data integrity of the model training and the clarity of the variable correspondence.

[0102] Step 2.5.3: Standardize the initial multivariate regression dataset and use the variance inflation factor method to diagnose multicollinearity, removing independent variables with variance inflation factors greater than a preset threshold to obtain the processed dataset.

[0103] This embodiment optimizes the quality of model input features, reduces linear dependencies between variables, and improves model stability and prediction accuracy by standardization and multicollinearity diagnosis to eliminate redundant variables.

[0104] Step 2.5.4: Based on the processed dataset, the molecular structure parameters of the interface modifiers are screened using the stepwise regression method with the p-value of the statistical test being less than a preset threshold as the criterion, and an initial quantitative correlation model is established.

[0105] This embodiment uses stepwise regression to screen key structural parameters, which can identify independent variables that have a significant impact on interface performance, simplify model complexity, and enhance the interpretability and engineering application value of the model.

[0106] Step 2.5.5: The prediction accuracy of the initial quantitative correlation model is evaluated using the k-fold cross-validation method to obtain the evaluation results. Based on the evaluation results, the coefficients of the initial quantitative correlation model are regularized and optimized using the ridge regression algorithm to obtain the quantitative correlation model between the molecular structure parameters of the interface modifier and the interface binding energy and the interface thermal conductivity.

[0107] This embodiment optimizes the model coefficients through k-fold cross-validation and ridge regression, which can evaluate the model's generalization ability and prevent overfitting, improve the predictive robustness of the quantitative association model, and ensure the model's reliability on unknown data.

[0108] Step 2.6: Based on the quantitative correlation model, a multi-objective optimization algorithm is used to iteratively optimize the molecular structure parameters of the interface modifier with the goal of simultaneously maximizing the interfacial binding energy and interfacial thermal conductivity, so as to obtain the optimal interfacial parameters.

[0109] This embodiment uses a quantitative correlation model for multi-objective optimization, which can simultaneously maximize the interfacial bonding energy and interfacial thermal conductivity, achieve synergistic improvement of interfacial performance, obtain the optimal combination of interfacial parameters, and significantly improve the overall performance of high thermal conductivity epoxy resin composite materials.

[0110] Step 3: Based on the optimal interface parameters, the mesoscopic structural characteristic parameters are obtained by simulating the dispersion process of the filler and the evolution process of the interface layer using the phase field method.

[0111] Step 3.1: Based on the optimal interface parameters, establish a three-phase phase field model including the epoxy resin matrix phase, filler phase, and interface modifier phase, and define three sets of sequence parameters to characterize the distribution state of the epoxy resin matrix phase, filler phase, and interface modifier phase, respectively.

[0112] This embodiment constructs a three-phase phase field model including an epoxy resin matrix phase, a filler phase, and an interface modifier phase, and defines three sets of sequence parameters, which can accurately quantify the spatial distribution state of each phase and the evolution characteristics of the phase interface at the mesoscale.

[0113] Step 3.2: Based on the interface binding energy and interface thermal conductivity in the optimal interface parameters, set the interface energy parameters and gradient energy coefficients between each phase in the three-phase phase field model.

[0114] This embodiment sets the interface energy parameters and gradient energy coefficient of the three-phase phase field model based on the interface binding energy and interface thermal conductivity in the optimal interface parameters. This enables the coupling of parameters between the microscopic interface performance and the mesoscopic phase field model, improves the model's prediction accuracy of interface heat transfer efficiency and adsorption intensity, and provides accurate physical boundary conditions for the packing dispersion kinetics and interface evolution process.

[0115] Step 3.3: Run the three-phase phase field model until the system reaches a dynamic equilibrium state.

[0116] This embodiment, by running a three-phase phase field model to a dynamic equilibrium state, can dynamically capture the entire process of packing dispersion-migration-agglomeration in a shear flow field and the evolution law of interface modifier adsorption-film formation-interfacial layer. It reflects the direct influence of actual processing parameters (such as shear rate and matrix viscosity) on the formation of mesoscopic structures, and enhances the correlation between simulation results and real processing.

[0117] Step 3.3.1: Introduce a convection term describing the shear flow field into the three-phase phase field model. By coupling the flow field control equations and the phase field evolution equations, simulate the dispersion, migration, and agglomeration kinetics of the filler in the epoxy resin matrix until the system reaches a dynamic equilibrium state, including: Based on the actual processing parameters, the flow field parameters in the three-phase phase field model are initialized, including the shear rate, viscosity and density of the epoxy resin matrix; Establish the coupled flow field control equations and phase field evolution equations; A convection term describing the shear flow field is introduced into the phase field evolution equation. The convection term is composed of the product of the flow field velocity vector and the order parameter gradient, and is used to describe the influence of the flow field on the packing transport and interface evolution. The coupled equations are spatially discretized using the finite difference method, and time integration is performed using an implicit-explicit hybrid time-progression method. Within each time step, the flow field control equations are first solved to obtain the updated flow field velocity distribution. Then, the flow field velocity distribution is substituted as a known quantity into the phase field evolution equations to solve for the order parameters of each time step. Based on the sequence parameters at each time step, the dispersion, migration and aggregation kinetics of fillers in the epoxy resin matrix are simulated through an iterative method until the system reaches a dynamic equilibrium state.

[0118] This embodiment, by coupling the flow field control equation and the phase field evolution equation and introducing a convection term, can accurately simulate the transport dynamics of the filler in the epoxy resin matrix, quantify the influence mechanism of the shear flow field on the spatial distribution, orientation, and agglomeration behavior of the filler, provide theoretical guidance for optimizing the filler dispersion process and suppressing agglomeration defects, and improve the uniformity of the filler network and the continuity of the heat conduction path.

[0119] Step 3.3.2: Set the interface gradient parameters according to the optimal interface parameters, and simulate the adsorption process of the interface modifier on the filler surface and the formation and evolution process of the interface layer by solving the interface evolution equation until the system reaches a dynamic equilibrium state, including: Based on the interfacial binding energy in the optimal interfacial parameters, the adsorption energy parameters of the interfacial modifier on the filler surface are determined, and the interfacial gradient energy coefficient in the three-phase phase field model is set based on the adsorption energy parameters. Initialize the concentration distribution field of the interface modifier in the epoxy resin matrix and set adsorption boundary conditions on the filler surface; A non-conservative Allen-Cahn equation describing the evolution of the interface layer is established. The non-conservative Allen-Cahn equation includes an interface gradient term, a double-well potential function term, and a convection transport term. Based on the interfacial gradient energy coefficient and the adsorption energy parameter, the interfacial evolution equation is solved to simulate the adsorption process of the interfacial modifier on the filler surface and the formation and evolution process of the interfacial layer until the system reaches a dynamic equilibrium state.

[0120] This embodiment sets the adsorption energy parameters based on the interfacial binding energy and solves the non-conservative Allen-Cahn equation, which can dynamically simulate the adsorption process of interfacial modifiers on the filler surface and the formation-evolution-stabilization process of the interfacial layer. It reveals the microscopic control mechanism of interfacial layer thickness distribution, uniformity and interfacial thermal conductivity, and provides a quantitative basis for the molecular design of interfacial modifiers and the optimization of interfacial performance.

[0121] Step 3.4: Extract the spatial distribution, orientation distribution, packing network shape, and interface layer thickness distribution of the packing under dynamic equilibrium state as simulation results.

[0122] This embodiment extracts simulation results such as the spatial distribution, orientation distribution, packing network shape, and interface layer thickness distribution of the packing under dynamic equilibrium conditions, which can form a direct observation dataset of mesoscopic structural characteristic parameters.

[0123] Step 3.5: Based on the simulation results, quantitatively calculate the permeation threshold, thermal conductivity path density, average thickness of the interface layer and its distribution uniformity of the packing network to form a set of mesoscopic structural feature parameters that include packing distribution characteristics and interface layer characteristics.

[0124] This embodiment quantifies the permeation threshold, thermal conductivity path density, average thickness of the interface layer, and its distribution uniformity of the filler network. This allows for the formation of a set of mesoscopic structural characteristic parameters that include both filler distribution characteristics and interface layer characteristics. This provides crucial mesoscopic structural inputs for finite element analysis to predict macroscopic properties such as thermal conductivity and electrical strength, achieving a precise correlation mapping from mesoscopic structure to macroscopic performance and improving the accuracy and reliability of composite material performance prediction.

[0125] Step 4: Based on the mesoscopic structural characteristic parameters, predict the macroscopic thermal conductivity, electrical strength and flexural strength of the high thermal conductivity epoxy resin composite material through finite element analysis to obtain the predicted performance values.

[0126] Step 4.1: Based on the filler spatial distribution, orientation distribution and filler network shape in the mesoscopic structural characteristic parameters, reconstruct the three-dimensional geometric model of the high thermal conductivity epoxy resin composite material.

[0127] This embodiment reconstructs a three-dimensional geometric model based on mesoscopic structural feature parameters, which can accurately restore the microstructural features of the packing's spatial distribution, orientation arrangement, and network morphology, providing high-fidelity geometric input for macroscopic performance prediction and improving the accuracy of performance predictions such as thermal conductivity and electrical strength.

[0128] Step 4.2: Based on the filler type and interface layer thickness distribution, assign corresponding material properties to the filler phase, interface layer and epoxy resin matrix phase, including thermal conductivity, dielectric constant, electrical conductivity and elastic modulus.

[0129] This embodiment achieves a quantitative correlation between mesoscopic structural parameters and macroscopic physical properties by assigning matching material properties to the filler phase, interface layer, and matrix. This ensures that properties such as thermal conductivity and dielectric constant are consistent with actual material characteristics, thereby enhancing the engineering reliability of finite element analysis.

[0130] Step 4.3: Mesh the three-dimensional geometric model and establish finite element models for the heat conduction control equation, the electrostatic conduction control equation, and the linear elasticity control equation, respectively.

[0131] This embodiment meshes the 3D model and establishes a multiphysics finite element model, which enables numerical solutions to thermo-mechanical-electric multi-field coupling problems. It provides a structured computational framework for the coordinated prediction of macroscopic thermal conductivity, electrical strength and bending strength, and improves the prediction efficiency of multiple performance indicators.

[0132] Step 4.4: Apply temperature boundary conditions to the finite element model of heat conduction, solve for the temperature field distribution, and calculate the macroscopic thermal conductivity of the high thermal conductivity epoxy resin composite material based on Fourier's law.

[0133] This embodiment solves the heat conduction equation by applying temperature boundary conditions, which can accurately quantify the macroscopic thermal conductivity of composite materials, reveal the influence of filler distribution and interfacial thermal conductivity on the overall heat transfer efficiency, and guide the optimization design of heat conduction paths.

[0134] Step 4.5: Apply voltage boundary conditions to the finite element model of electrostatic conduction, solve for the electric field distribution, and predict the electrical strength of the high thermal conductivity epoxy resin composite material based on the extreme values ​​of the electric field strength.

[0135] This embodiment solves the electrostatic conduction equation based on voltage boundary conditions, which can predict the extreme values ​​of the electrical strength of the material, evaluate the dispersion state of the filler and the regulatory effect of the interface layer on the electric field distribution, and provide a quantitative basis for optimizing insulation performance.

[0136] Step 4.6: Apply displacement and stress boundary conditions to the finite element model of linear elasticity, solve for the stress-strain field, and predict the flexural strength of the high thermal conductivity epoxy resin composite material based on the maximum stress criterion.

[0137] In this embodiment, displacement and stress boundary conditions are applied to the finite element model of linear elasticity to obtain the stress-strain field. This quantifies the influence of the filler network morphology and interface bonding strength on mechanical properties, thereby improving the accuracy of bending performance design.

[0138] Step 4.7: Use the macroscopic thermal conductivity, electrical strength, and flexural strength as performance prediction values.

[0139] In this embodiment, the macroscopic thermal conductivity, electrical strength, and flexural strength are used as performance prediction values, which can form a quantitative evaluation system of multiple performance indicators, providing clear optimization targets and data benchmarks for subsequent performance regulation.

[0140] Step 5: Adjust the filler volume fraction and interface modification strength to regulate performance based on the difference between the predicted performance value and the target performance value.

[0141] Step 5.1: Calculate the relative error between the predicted performance value and the target performance value, wherein the predicted performance value includes the predicted macroscopic thermal conductivity, the predicted electrical strength, and the predicted flexural strength.

[0142] This embodiment quantifies the prediction deviation of each performance indicator by calculating the relative error between the predicted performance value and the target value, providing data support for optimization and ensuring the targeted nature of performance regulation.

[0143] Step 5.2: Establish a multi-objective optimization function based on the relative error. The multi-objective optimization function aims to minimize the weighted sum of the relative errors of the predicted macroscopic thermal conductivity, predicted electrical strength, and predicted bending strength.

[0144] This embodiment constructs a multi-objective optimization function to balance the competing relationships between thermal conductivity, electrical strength, and flexural strength. By minimizing the weighted error, it achieves synergistic optimization of multiple performance indicators and improves the efficiency of overall performance control.

[0145] Step 5.3: Identify the key control parameters for macroscopic thermal conductivity, electrical strength and flexural strength. The key control parameters include filler volume fraction and interface modification strength, wherein the interface modification strength is characterized by the functional group density and molecular chain length of the interface modifier.

[0146] This embodiment identifies filler volume fraction and interface modification intensity as key control parameters, allowing for focused optimization of core variables, reducing the blindness of parameter adjustments, and improving the convergence speed and accuracy of the optimization process.

[0147] Step 5.4: Based on the gradient descent algorithm or genetic algorithm, iteratively optimize the filler volume fraction and interface modification intensity within the preset parameter adjustment range, wherein the interface modification intensity is characterized by the functional group density and molecular chain length of the interface modifier.

[0148] This embodiment uses gradient descent or genetic algorithm for iterative optimization, which can efficiently search the parameter space, dynamically adjust the filler volume fraction and interface modification intensity, and realize closed-loop control of the entire process from mesoscopic structure to macroscopic performance.

[0149] Step 5.5: When the value of the multi-objective optimization function converges within the preset tolerance range, or reaches the maximum number of iterations, stop the optimization process and obtain the optimal combination of filler volume fraction and interface modification strength.

[0150] This embodiment stops optimization by limiting the convergence of the optimization function or the number of iterations, which ensures the engineering feasibility of the optimal parameter combination, avoids performance fluctuations caused by over-optimization, and improves the stability of the control results.

[0151] Step 5.6: Adjust the performance based on the optimal combination of filler volume fraction and interfacial modification strength.

[0152] This embodiment uses optimal parameter combinations for performance regulation, which can achieve precise matching of multiple performance indicators of high thermal conductivity epoxy resin composites in terms of heat, force, and electricity, meet preset performance target values, and enhance the engineering application value of the comprehensive performance of high thermal conductivity epoxy resin composites.

[0153] Compared with the prior art, the beneficial effects of this embodiment are as follows: 1. A cross-scale inverse decomposition system of "macroscopic performance targets - mesoscopic structural features - microscopic interface parameters" was constructed. Combining high-throughput computing, molecular dynamics simulation and other technologies, it overcomes the shortcomings of existing technologies, such as lack of multi-scale collaborative optimization and weak generalization ability of empirical models. It realizes the precise mapping from performance targets to controllable parameters, providing a systematic solution for the independent design of core materials. Ultimately, it enables high thermal conductivity epoxy resin composites to stably achieve the preset performance targets of thermal conductivity >1W / mK; electrical strength >20MV / m; and flexural strength >105MPa, significantly improving the controllability of material properties and the design success rate. 2. A quantitative correlation model was established between the molecular structure parameters of the interface modifier and the interfacial binding energy and interfacial thermal conductivity. Through multi-objective optimization, the simultaneous maximization of the two types of interfacial properties was achieved, overcoming the shortcomings of existing technologies that lack quantitative basis for interface matching and make it difficult to accurately screen combination schemes. It effectively broke through the industry bottleneck that the thermal conductivity of traditional epoxy resin is only >0.6W / mK, and provided core technical support for the synergistic improvement of thermal conductivity, mechanical properties and insulation properties. 3. By accurately predicting macroscopic performance through finite element analysis and constructing a closed-loop control circuit based on performance deviation, the filler volume fraction and interface modification intensity are dynamically adjusted. This overcomes the shortcomings of existing technologies, such as the disconnect between mesoscopic simulation and actual processing and the lack of closed-loop control. It ensures that the final material performance accurately matches the preset target, while significantly reducing dependence on imported core technologies and materials. This lays a solid foundation for cost reduction, efficiency improvement, and technological self-reliance in large-capacity dry-type transformers.

[0154] Further, in step S2, based on the application conditions and expected performance indicators of the high thermal conductivity epoxy resin composite material in dry-type transformers, a calibration sample containing microcracks of a predetermined morphology is prepared, specifically: Based on the application conditions and expected performance indicators, the preset morphological parameters of microcracks are determined through simulation analysis. The expected performance indicators include thermal conductivity, volume resistivity, flexural strength and heat deformation temperature. The preset morphological parameters include the number of microcracks, the length of microcracks, the width of microcracks, the depth of microcracks and their spatial distribution coordinates. The calibration sample to be prepared is fixed according to its actual assembly orientation in the dry-type transformer winding; By applying controlled mechanical stress or thermal shock treatment, microcracks consistent with the preset morphological parameters are introduced into the fixed calibration sample to obtain the calibration sample containing microcracks.

[0155] Further, in step S2, the calibration sample is scanned with multi-frequency electromagnetic waves to acquire and process the electromagnetic wave response signal, generating the electromagnetic wave response characteristic spectrum of the calibration sample. Specifically: Based on the material properties of the calibration sample and the preset morphological parameters of the microcracks, the scanning parameters of the multi-frequency electromagnetic waves are set. The material properties of the calibration samples include thermal conductivity, dielectric constant, and elastic modulus. The scanning parameters include a frequency range of 0.1THz to 2.7THz and a scanning resolution of not less than 6GHz; Based on the scanning parameters, a full-coverage scan is performed on the calibration sample fixed in the actual assembly position of the dry-type transformer winding, and the original electromagnetic wave response signal is acquired in real time. The original electromagnetic wave response signal is preprocessed to obtain a purified signal. The preprocessing includes wavelet threshold noise reduction and matched filtering enhancement. Extract characteristic parameters, including characteristic frequency, characteristic amplitude, reflection coefficient, and projection coefficient, from the purified signal; The characteristic parameters are integrated and normalized to generate the electromagnetic wave response characteristic spectrum of the calibration sample.

[0156] Further, the step of integrating and normalizing the characteristic parameters to generate the electromagnetic response characteristic spectrum of the calibration sample includes: Based on the characteristic frequency and the characteristic amplitude, a clustering analysis method is used to perform pattern recognition to obtain a cluster of characteristic parameters representing different microcrack types; Based on the cluster of characteristic parameters, a multi-level micro-defect information superposition algorithm is used to perform spatial domain fusion processing on the reflection coefficient and projection coefficient to obtain enhanced defect spatial distribution characteristics. Based on the cluster of characteristic parameters and the enhanced spatial distribution characteristics of defects, a quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters is constructed. Based on the quantitative mapping relationship, the characteristic parameters of the entire frequency band are normalized and encoded to generate an electromagnetic wave response characteristic spectrum containing characteristic frequency bands, characteristic amplitudes, and spatial distribution coordinates.

[0157] Further, the step of using a multi-level micro-defect information overlay algorithm to perform spatial domain fusion processing on the reflection coefficient and projection coefficient based on the feature parameter cluster to obtain enhanced defect spatial distribution features includes: Based on the cluster of characteristic parameters, the amplitude gradient distribution of the reflection coefficient and the phase delay distribution of the projection coefficient are adaptively divided into multiple data subsets according to the scanning space coordinates. The number of the subsets is positively correlated with the density of the microcrack spatial distribution coordinates. Based on the multiple hierarchical data subsets, the local amplitude maxima of the reflection coefficients and the average phase delay of the projection coefficients within each hierarchical level are extracted to construct the feature vector of the current level. Principal component analysis was used to reduce the dimensionality of the feature vectors at each level and fuse them to obtain the defect probability distribution map corresponding to each level. Based on spatial coordinates, the defect probability distribution maps corresponding to each level are weighted and superimposed to obtain the superimposed defect probability distribution map. The superimposed defect probability distribution map is processed by morphological opening operation to filter out noise and connect adjacent defect regions to obtain enhanced defect spatial distribution features.

[0158] Further, in step S2, based on the electromagnetic wave response characteristic spectrum, a clustering analysis and multi-level micro-defect information superposition algorithm are used to extract the quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters, and an electromagnetic feature fingerprint database is established based on the quantitative mapping relationship. Specifically: Based on the electromagnetic wave response characteristic spectrum, the K-means++ clustering algorithm is used to perform unsupervised classification of the characteristic frequency bands and characteristic amplitudes, resulting in an initial mapping relationship lookup table with the cluster center as the index and associated with the mean length and mean width of the corresponding microcracks. Based on the spatial distribution coordinates in the electromagnetic wave response characteristic spectrum, a multi-level micro-defect information superposition algorithm is used to extract the spatial distribution characteristics of defects. Based on the aforementioned spatial distribution characteristics of defects, the correlation features between the depth of microcracks and their spatial distribution coordinates are extracted by Gaussian mixture model clustering. The number of components in the Gaussian mixture model clustering is adaptively determined according to the number of microcracks in the preset morphological parameters. Based on the initial mapping lookup table and the associated features, a microcrack morphology-electromagnetic feature parameter mapping relationship matrix is ​​constructed, which uses feature frequency and feature amplitude as parameters and associates microcrack length, microcrack width, microcrack depth and spatial distribution coordinates. Based on the microcrack morphology-electromagnetic feature parameter mapping matrix, the feature parameters in the electromagnetic wave response feature spectrum are normalized and encoded to establish the electromagnetic feature fingerprint database.

[0159] Furthermore, the step of extracting the correlation features between the depth of microcracks and their spatial distribution coordinates through Gaussian mixture model clustering based on the spatial distribution characteristics of the defects includes: The spatial distribution characteristics of the defects are used as the input feature vector for Gaussian mixture model clustering; The number of components for Gaussian mixture model clustering is adaptively determined based on the number of microcracks in the preset morphological parameters. The expectation-maximization algorithm is used to estimate the parameters and iteratively optimize the Gaussian mixture model until the Gaussian mixture model converges. Based on the converged Gaussian mixture model, the posterior probability of each microcrack data point belonging to each Gaussian component is calculated. Based on the maximum a posteriori probability principle, each microcrack data point is assigned to the corresponding Gaussian component to complete defect clustering; Extract the mean vector and covariance matrix of each Gaussian component, where the mean vector represents the typical depth and spatial coordinates of the current type of microcrack, and the covariance matrix represents the distribution pattern of the typical depth and spatial coordinates of the current type of microcrack. Based on the mean vector and covariance matrix of each Gaussian component, a three-dimensional correlation feature matrix between the depth of the microcrack and its spatial distribution coordinates is constructed, which serves as the correlation feature between the depth of the microcrack and its spatial distribution coordinates.

[0160] Furthermore, in step S2, during the accelerated aging test simulating the operating conditions of a dry-type transformer, the high thermal conductivity epoxy resin composite material sample undergoes multi-stage electromagnetic wave scanning, and electromagnetic wave scanning data is generated by encoding according to a time sequence. Specifically: The high thermal conductivity epoxy resin composite material sample was fixed in the aging test device according to the actual assembly orientation of the dry transformer winding. Based on the actual operating conditions of the dry-type transformer, the environmental parameters for the aging test are set, including: a temperature cycling range of -40°C to +120°C, a relative humidity range of 20% to 95%, a mechanical vibration frequency of 10Hz to 200Hz and an amplitude of 0.5g to 5g, and a voltage level of 0.5 times to 1.5 times the rated voltage. Accelerated aging tests are conducted according to a preset aging cycle, which includes multiple consecutive aging stages. Each aging stage includes a heating sub-stage, a heat preservation sub-stage, a cooling sub-stage, and a voltage loading sub-stage. After each aging stage, the high thermal conductivity epoxy resin composite material sample is scanned using the multi-frequency electromagnetic wave scanning parameters to collect the original electromagnetic wave response signal. The collected raw electromagnetic wave response signal is preprocessed to obtain a purified signal. The preprocessing includes wavelet threshold denoising and matched filtering enhancement. From the purified signal, feature parameters including characteristic frequency, characteristic amplitude, reflection coefficient and projection coefficient are extracted to obtain the feature parameters of each stage; According to the aging time series, the characteristic parameters of each stage are encoded in time series to generate electromagnetic wave scanning data.

[0161] Furthermore, in step S2, a deep learning model is used to detect and evaluate the electromagnetic feature fingerprint database and electromagnetic wave scanning data to obtain quantitative detection results and quantitative evaluation results, specifically: Construct a deep learning model that includes convolutional neural network branches and long short-term memory network branches; The electromagnetic fingerprint database is input into the branch of the convolutional neural network to extract spatial features and obtain static fingerprint features. The electromagnetic wave scanning data is input into the long short-term memory network branch to extract dynamic temporal features. A cross-modal attention fusion module is used to adaptively weight and fuse static fingerprint features and dynamic temporal features to obtain a fused feature vector; The fused feature vector is input into a fully connected classification layer to classify the defect type and regress the spatial location, and output the type and location of defects in the high thermal conductivity epoxy resin composite sample. The fused feature vectors are simultaneously input into a regression layer to quantitatively predict the defect size and output the size of the defect in the high thermal conductivity epoxy resin composite sample. The dynamic time series characteristics are input into a time series analysis layer to evaluate the aging degree of high thermal conductivity epoxy resin composite material samples and predict the remaining life. The predicted values ​​of aging degree and remaining life of high thermal conductivity epoxy resin composite material samples are output. The quantitative detection results are obtained by integrating the outputs of the classification layer and the regression layer. The quantitative evaluation results are obtained by integrating the output of the time series analysis layer.

[0162] Furthermore, step S2 also includes: generating a health status report for the high thermal conductivity epoxy resin composite material based on the quantitative detection results and the quantitative assessment results, wherein the health status report includes: A planar schematic diagram of a high thermal conductivity epoxy resin composite sample, including the type, location, and size of defects. Aging degree of high thermal conductivity epoxy resin composite material sample as a function of time; Predicted remaining life of high thermal conductivity epoxy resin composite material specimens.

[0163] For example: This embodiment introduces a detection and evaluation method based on high thermal conductivity epoxy resin composite materials, including: Step 1: Based on the application conditions and expected performance indicators of high thermal conductivity epoxy resin composite materials in dry-type transformers, prepare calibration samples containing microcracks of a predetermined morphology.

[0164] By preparing calibration samples containing microcracks of predetermined morphology based on the application conditions and expected performance indicators of high thermal conductivity epoxy resin composites in dry-type transformers, subsequent testing and evaluation are grounded in high-fidelity simulation. Simulation analysis was used to determine the predetermined morphological parameters of the microcracks. After fixing the calibration samples according to their actual assembly orientation, microcracks were introduced, ensuring that the defect morphology and spatial distribution in the calibration samples accurately reflect the damage modes most likely to occur in actual operation. This lays a reliable physical foundation for establishing a high-precision quantitative mapping relationship and effectively overcomes evaluation biases caused by discrepancies between the calibration samples and actual operating conditions.

[0165] Step 2: Scan the calibration sample with multi-frequency electromagnetic waves to acquire and process the electromagnetic wave response signal to generate the electromagnetic wave response characteristic spectrum of the calibration sample.

[0166] By employing multi-frequency electromagnetic wave scanning and setting a frequency range covering 0.1 THz to 2.7 THz with a scanning resolution of no less than 6 GHz, a broadband electromagnetic response related to microcracks within the calibrated sample can be excited and captured. Subsequent signal preprocessing, including wavelet thresholding and matched filtering enhancement, significantly improves signal quality. By extracting characteristic parameters such as characteristic frequencies, characteristic amplitudes, reflection coefficients, and projection coefficients from the purified signal and integrating them to generate an electromagnetic wave response characteristic spectrum, microscopic physical defects are transformed into quantifiable and analyzable electromagnetic characteristic data, providing a rich and accurate information source for subsequent quantitative identification.

[0167] Step 3: Based on the electromagnetic wave response characteristic spectrum, cluster analysis and multi-level micro-defect information superposition algorithm are used to extract the quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters, and an electromagnetic feature fingerprint database is established based on the quantitative mapping relationship.

[0168] Based on the electromagnetic wave response characteristic spectrum, a cluster of characteristic parameters is obtained by using cluster analysis. Then, a multi-level micro-defect information superposition algorithm is used to perform spatial domain fusion processing on the reflection coefficient and projection coefficient to obtain enhanced defect spatial distribution characteristics. In this way, a quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters is constructed. The electromagnetic feature fingerprint database established based on the quantitative mapping relationship is specifically represented as a microcrack morphology-electromagnetic characteristic parameter mapping relationship matrix. It accurately associates discrete electromagnetic features with specific microcrack length, width, depth and spatial distribution coordinates, forming a knowledge base that can be used for rapid comparison and retrieval, realizing accurate and quantitative inversion from electromagnetic signal to defect morphology.

[0169] Step 4: In the accelerated aging test simulating the operating conditions of a dry-type transformer, the high thermal conductivity epoxy resin composite material sample was subjected to multi-stage electromagnetic wave scanning, and the electromagnetic wave scanning data was generated by encoding the data according to the time sequence.

[0170] By setting environmental parameters including temperature cycling, humidity, mechanical vibration, and voltage loading in accelerated aging tests simulating the operating conditions of dry-type transformers, and performing multi-stage electromagnetic wave scanning on high thermal conductivity epoxy resin composite material samples, the entire life cycle aging process of materials in actual operation can be simulated. After each aging stage, signals are collected and processed, characteristic parameters are extracted, and electromagnetic wave scanning data are generated by encoding according to the aging time series. This yields a time-series dataset that dynamically reflects the material performance degradation process, providing crucial data support for assessing the aging degree of materials and predicting their remaining lifespan.

[0171] Step 5: Use a deep learning model to detect and evaluate the electromagnetic feature fingerprint database and the electromagnetic wave scanning data to obtain quantitative detection results and quantitative evaluation results.

[0172] In this embodiment, the quantitative detection results include the type, location, and size of defects in the high thermal conductivity epoxy resin composite material sample, and the quantitative evaluation results include the aging degree of the high thermal conductivity epoxy resin composite material sample and the predicted value of its remaining life.

[0173] By constructing a deep learning model incorporating branches of convolutional neural networks and long short-term memory networks, and utilizing a cross-modal attention fusion module to adaptively fuse static fingerprint features extracted from an electromagnetic fingerprint database and dynamic temporal features extracted from electromagnetic wave scanning data, the synergistic advantages of prior knowledge and real-time monitoring data are fully leveraged. The model outputs the type, location, and size of defects through fully connected classification and regression layers, and predicts aging degree and remaining lifespan through a time series analysis layer. Ultimately, it achieves a comprehensive, automated, and quantitative assessment of the internal state of high thermal conductivity epoxy resin composite materials, from static defect detection to dynamic lifespan prediction, significantly improving the accuracy, efficiency, and reliability of the detection and assessment.

[0174] The implementation steps of a detection and evaluation method based on high thermal conductivity epoxy resin composite materials include: Step 1: Based on the application conditions and expected performance indicators of high thermal conductivity epoxy resin composite materials in dry-type transformers, prepare calibration samples containing microcracks of a predetermined morphology.

[0175] Step 1.1: Based on the application conditions and expected performance indicators, determine the preset morphological parameters of the microcracks through simulation analysis.

[0176] In this embodiment, the expected performance indicators include thermal conductivity, volume resistivity, flexural strength, and heat distortion temperature, and the preset morphological parameters include the number of microcracks, the length of microcracks, the width of microcracks, the depth of microcracks, and the spatial distribution coordinates.

[0177] This embodiment determines the preset morphological parameters of microcracks through simulation analysis based on the application conditions and expected performance indicators, giving the preparation of calibration samples a clear target orientation and scientific basis. It can pre-simulate and reproduce the most sensitive and easily occurring defect morphologies in materials during actual operation, ensuring that the prepared calibration samples accurately cover key failure modes. This lays a solid foundation for establishing high-precision detection and evaluation standards and effectively avoids the problem of detection model failure caused by blind or unrealistic defect morphology settings.

[0178] Step 1.2: Fix the calibration sample to be prepared according to its actual assembly orientation in the dry-type transformer winding.

[0179] This embodiment effectively simulates the stress state and constraint conditions of materials in real products by fixing the calibration sample to be prepared according to its actual assembly orientation in the winding of a dry-type transformer. This step ensures that when microcracks are subsequently introduced, the path and direction of their generation and propagation can highly reproduce the damage behavior under actual working conditions. This makes the spatial distribution and morphological characteristics of microcracks in the calibration sample more representative and realistic, significantly improving the consistency between the calibration sample and the actual object under test, and providing a crucial prerequisite for obtaining reliable electromagnetic fingerprints.

[0180] Step 1.3: By applying controlled mechanical stress or thermal shock treatment, microcracks consistent with the preset morphological parameters are introduced into the fixed calibration sample to obtain the calibration sample containing microcracks.

[0181] This embodiment introduces microcracks through controlled mechanical stress loading or thermal shock treatment, achieving precise control over the defect generation process. It ensures that the preset morphological parameters of the microcracks introduced into the fixed calibration sample are consistent with the simulation design target, thereby obtaining a calibration sample containing microcracks with known and controllable defect morphology.

[0182] Step 2: Scan the calibration sample with multi-frequency electromagnetic waves to acquire and process the electromagnetic wave response signal to generate the electromagnetic wave response characteristic spectrum of the calibration sample.

[0183] Step 2.1: Set the scanning parameters of the multi-frequency electromagnetic wave according to the material properties of the calibration sample and the preset morphological parameters of the microcrack.

[0184] In this embodiment, the material properties of the calibration sample include thermal conductivity, dielectric constant, and elastic modulus, and the scanning parameters include a frequency range of 0.1 THz to 2.7 THz and a scanning resolution of not less than 6 GHz.

[0185] The above scanning parameters are closely related to material properties and defect morphology, as detailed below: Dielectric constant dominates frequency band planning: The dielectric constant of a material directly determines the propagation characteristics of terahertz waves. For composite materials with high dielectric constants, such as epoxy resin, high-frequency signal attenuation is significant. Therefore, the frequency band is set from 0.1THz to 2.7THz, achieving optimal synergy between the deep penetration capability in the low-frequency band (0.1–1THz) and the high-resolution imaging advantage in the high-frequency band (1–2.7THz).

[0186] The relationship between thermal conductivity and elastic modulus and the system dynamic range: Thermal conductivity and elastic modulus indirectly reflect the uniformity and density of the material's microstructure, affecting the electromagnetic scattering intensity at defect interfaces. These properties provide a basis for setting a system dynamic range >60dB, ensuring that weak scattering signals induced by microcracks can be effectively extracted from the material's background response.

[0187] Defect morphology determines frequency resolution: the pre-defined morphological parameters of microcracks (such as width and depth) directly correspond to the characteristic frequencies of their electromagnetic response. To ensure accurate resolution of the minute spectral characteristics of defects ranging from micrometers to sub-millimeters, a high scanning resolution of at least 6 GHz is required. This lays the data foundation for subsequent quantitative inversion of the actual physical size of defects from the electromagnetic response characteristic spectrum.

[0188] This embodiment sets multi-frequency electromagnetic wave scanning parameters based on the material properties of the calibrated sample and the preset morphological parameters of the microcracks, enabling the electromagnetic wave scanning process to be optimized for specific material systems and target defects. The wide frequency range ensures coverage of excitation and response to microcracks of different sizes, while the high scanning resolution guarantees the ability to capture minute defect features, thus providing optimal detection conditions for achieving high-precision microcrack detection and identification.

[0189] Step 2.2: Based on the scanning parameters, perform a full-coverage scan on the calibration sample fixed in the actual assembly position of the dry-type transformer winding, and collect the original electromagnetic wave response signal in real time.

[0190] This embodiment performs a full-coverage scan of the calibration sample fixed in the actual assembly position of the dry-type transformer winding based on the aforementioned scanning parameters. This ensures that the collected electromagnetic wave data can fully reflect the overall electromagnetic characteristics of the calibration sample under actual working conditions. Real-time acquisition of the original electromagnetic wave response signal guarantees the synchronicity and completeness of data acquisition, providing a comprehensive and complete original data foundation for constructing an electromagnetic wave response characteristic spectrum that can accurately characterize the internal state of the material.

[0191] Step 2.3: Perform signal preprocessing on the original electromagnetic wave response signal to obtain a purified signal. The signal preprocessing includes wavelet threshold denoising and matched filtering enhancement.

[0192] This embodiment effectively suppresses environmental noise and system interference by performing signal preprocessing on the original electromagnetic wave response signal, including wavelet threshold denoising and matched filtering enhancement. At the same time, it highlights the effective signal components caused by microcracks, significantly improving the signal-to-noise ratio of the signal. This makes the purified signal on which subsequent feature extraction depends more pure and reliable, laying a key data quality foundation for accurately generating the electromagnetic wave response feature spectrum.

[0193] Step 2.4: Extract characteristic parameters, including characteristic frequency, characteristic amplitude, reflection coefficient and projection coefficient, from the purified signal.

[0194] This embodiment extracts characteristic parameters, including characteristic frequency, characteristic amplitude, reflection coefficient, and projection coefficient, from the purification signal, realizing the transformation from complex time-domain / frequency-domain signals to key indicators that can characterize the internal physical state of the material. These multi-dimensional characteristic parameters reflect the existence and influence of microcracks from different perspectives, providing a rich and representative set of data features for subsequent quantitative analysis and the establishment of mapping relationships.

[0195] Step 2.5: Integrate and normalize the characteristic parameters to generate the electromagnetic wave response characteristic spectrum of the calibration sample.

[0196] This embodiment integrates and normalizes the extracted feature parameters to generate the electromagnetic response feature spectrum of the calibrated sample. This step unifies the scattered multi-dimensional features into a standardized framework, eliminates the influence of dimensions, and forms a "digital fingerprint" that can uniquely characterize a specific microcrack morphology. This structured feature spectrum greatly facilitates subsequent pattern recognition, cluster analysis, and the establishment of quantitative mapping relationships, and is the core data carrier for achieving rapid and accurate defect detection and assessment.

[0197] Step 2.5.1: Based on the characteristic frequency and the characteristic amplitude, a clustering analysis method is used to perform pattern recognition to obtain a cluster of characteristic parameters representing different microcrack types.

[0198] This embodiment uses clustering analysis based on the characteristic frequency and characteristic amplitude to perform pattern recognition, obtaining feature parameter clusters characterizing different microcrack types. This automatically categorizes mixed electromagnetic feature signals into sets with clear physical meaning. This method does not require pre-defined classification rules; it spontaneously reveals the inherent differences and patterns in the electromagnetic responses of microcracks of different sizes and properties, achieving preliminary and automated differentiation of microcrack types. This provides a clear and reliable data structure foundation for subsequently constructing refined quantitative mapping relationships.

[0199] Step 2.5.2: Based on the feature parameter cluster, a multi-level micro-defect information superposition algorithm is used to perform spatial domain fusion processing on the reflection coefficient and projection coefficient to obtain enhanced defect spatial distribution features.

[0200] Based on the cluster of characteristic parameters, the amplitude gradient distribution of the reflection coefficient and the phase delay distribution of the projection coefficient are adaptively divided into multiple data subsets according to the scanning space coordinates. The number of the subsets is positively correlated with the density of the microcrack spatial distribution coordinates. Based on the multiple hierarchical data subsets, the local amplitude maxima of the reflection coefficients and the average phase delay of the projection coefficients within each hierarchical level are extracted to construct the feature vector of the current level. Principal component analysis was used to reduce the dimensionality of the feature vectors at each level and fuse them to obtain the defect probability distribution map corresponding to each level. Based on spatial coordinates, the defect probability distribution maps corresponding to each level are weighted and superimposed to obtain the superimposed defect probability distribution map. The superimposed defect probability distribution map is processed by morphological opening operation to filter out noise and connect adjacent defect regions to obtain enhanced defect spatial distribution features.

[0201] This embodiment employs a multi-level micro-defect information overlay algorithm to perform spatial domain fusion processing on the reflection coefficient and projection coefficient, achieving in-depth mining and enhancement of defect spatial distribution information. This method adaptively divides data into multiple levels of subsets and constructs feature vectors for each. Principal component analysis is then used for feature dimensionality reduction and fusion to obtain a defect probability distribution map. Finally, weighted overlay and morphological opening operations effectively fuse the amplitude information of the reflected signal and the phase information of the projected signal, significantly improving the signal-to-noise ratio. This allows for precise delineation of the complete contour and spatial location of microcracks, effectively filtering out noise and connecting adjacent defect regions, thereby obtaining clear, coherent, and accurate enhanced defect spatial distribution features.

[0202] Step 2.5.3: Based on the cluster of characteristic parameters and the enhanced spatial distribution characteristics of defects, construct a quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters.

[0203] This embodiment constructs a quantitative mapping relationship between microcrack morphology and electromagnetic feature parameters based on the feature parameter cluster and the enhanced defect spatial distribution characteristics. It successfully correlates and mathematically models the defect types identified in step 2.5.1 with the defect spatial morphology located in step 2.5.2, realizing a leap from "electromagnetic signal features" to "physical defect morphology parameters". It establishes a quantifiable and deterministic correspondence, providing a core conversion model and theoretical basis for upgrading electromagnetic detection technology from qualitative judgment to high-precision quantitative inversion.

[0204] Step 2.5.4: Normalize and encode the characteristic parameters of the entire frequency band according to the quantitative mapping relationship to generate an electromagnetic wave response characteristic spectrum containing characteristic frequency bands, characteristic amplitudes and spatial distribution coordinates.

[0205] This embodiment normalizes and encodes the characteristic parameters of the entire frequency band according to the quantitative mapping relationship, ultimately generating a structured electromagnetic wave response characteristic spectrum containing characteristic frequency bands, characteristic amplitudes, and spatial distribution coordinates. The electromagnetic wave response characteristic spectrum is not a simple list of original features, but rather a "defect ID card" calibrated and standardized by quantitative relationships. This provides a unified comparison benchmark and interpretation standard for electromagnetic characteristic data obtained from different samples and different scans, laying the final data output format foundation for rapid retrieval of electromagnetic feature fingerprint databases and the realization of accurate and efficient automated detection and evaluation.

[0206] Step 3: Based on the electromagnetic wave response characteristic spectrum, cluster analysis and multi-level micro-defect information superposition algorithm are used to extract the quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters, and an electromagnetic feature fingerprint database is established based on the quantitative mapping relationship.

[0207] Step 3.1: Based on the electromagnetic wave response characteristic spectrum, the K-means++ clustering algorithm is used to perform unsupervised classification of the characteristic frequency bands and characteristic amplitudes to obtain an initial mapping relationship lookup table with the cluster center as the index and associated with the mean length of the corresponding microcrack and the mean width of the microcrack.

[0208] This embodiment uses the K-means++ clustering algorithm to perform unsupervised classification of feature frequency bands and feature amplitudes based on the electromagnetic wave response feature spectrum. This automatically discovers the inherent grouping patterns of electromagnetic features without relying on prior knowledge for manual annotation. The method generates an initial mapping relationship lookup table indexed by cluster centers and associated with the mean length and mean width of corresponding microcracks. This establishes a preliminary quantitative correspondence between electromagnetic features and two-dimensional morphological parameters of microcracks, providing an important two-dimensional foundation for the subsequent construction of a complete mapping relationship matrix, while avoiding errors caused by subjective classification.

[0209] Step 3.2: Based on the spatial distribution coordinates in the electromagnetic wave response characteristic spectrum, extract the spatial distribution characteristics of defects using a multi-level micro-defect information superposition algorithm.

[0210] This embodiment extracts the spatial distribution features of defects by employing a multi-level micro-defect information superposition algorithm based on the spatial distribution coordinates in the electromagnetic wave response characteristic spectrum. This fully utilizes the spatial information obtained from electromagnetic wave scanning to deeply explore the distribution patterns of microcracks within the material. This step enhances the characterization of the spatial distribution characteristics of the microcrack population, providing a high-quality spatial distribution data foundation for the next step of accurately correlating the depth of microcracks, a key three-dimensional morphological parameter.

[0211] Step 3.3: Based on the spatial distribution characteristics of the defects, the correlation features between the depth of the microcracks and the spatial distribution coordinates are extracted by Gaussian mixture model clustering.

[0212] This embodiment achieves a complete characterization of the three-dimensional morphology of microcracks by extracting the correlation features between the depth and spatial distribution coordinates of microcracks through Gaussian mixture model clustering based on the spatial distribution characteristics of the defects. This method effectively models the complex probability distribution relationship between microcrack depth and spatial location, thereby organically fusing depth information with spatial coordinates. This provides crucial three-dimensional correlation features for ultimately establishing a full-dimensional quantitative mapping relationship including depth parameters.

[0213] Step 3.3.1: Use the spatial distribution features of the defects as the input feature vector for Gaussian mixture model clustering.

[0214] Step 3.3.2: Based on the number of microcracks in the preset morphological parameters, adaptively determine the number of components for Gaussian mixture model clustering.

[0215] Step 3.3.3: Use the expectation-maximization algorithm to estimate the parameters and iteratively optimize the Gaussian mixture model until the Gaussian mixture model converges.

[0216] Step 3.3.4: Based on the converged Gaussian mixture model, calculate the posterior probability of each microcrack data point belonging to each Gaussian component.

[0217] Step 3.3.5: Based on the maximum a posteriori probability principle, assign each microcrack data point to the corresponding Gaussian component to complete defect clustering.

[0218] Step 3.3.6: Extract the mean vector and covariance matrix of each Gaussian component, where the mean vector represents the typical depth and spatial coordinates of the current type of microcrack, and the covariance matrix represents the distribution pattern of the typical depth and spatial coordinates of the current type of microcrack.

[0219] Step 3.3.7: Based on the mean vector and covariance matrix of each Gaussian component, construct a three-dimensional correlation feature matrix between the depth of the microcrack and its spatial distribution coordinates, as the correlation feature between the depth of the microcrack and its spatial distribution coordinates.

[0220] In this embodiment, the number of components in the Gaussian mixture model clustering is adaptively determined based on the number of microcracks in the preset morphological parameters.

[0221] This embodiment ensures the applicability of the data input by using the spatial distribution characteristics of defects as the input feature vector; adaptively determining the number of components based on the number of microcracks ensures that the model complexity matches the actual problem; using the expectation-maximization algorithm for parameter estimation and iterative optimization ensures that the model converges to the optimal solution; calculating the posterior probability and allocating data points according to the maximum a posteriori probability principle achieves accurate defect clustering; finally, extracting the mean vector and covariance matrix and constructing a three-dimensional correlation feature matrix not only quantifies the typical depth and location of various microcracks, but also reveals their statistical distribution law, thus comprehensively and accurately establishing a probabilistic correlation model between microcrack depth and spatial distribution coordinates, laying a solid theoretical foundation for high-precision three-dimensional reconstruction and quantitative evaluation of defects.

[0222] Step 3.4: Based on the initial mapping relationship lookup table and the associated features, construct a microcrack morphology-electromagnetic feature parameter mapping relationship matrix with feature frequency and feature amplitude as parameters, and associated with microcrack length, microcrack width, microcrack depth and spatial distribution coordinates.

[0223] This embodiment constructs a microcrack morphology-electromagnetic feature parameter mapping matrix by querying the initial mapping relationship table and the associated features. The matrix uses feature frequency and feature amplitude as parameters and associates microcrack length, microcrack width, microcrack depth, and spatial distribution coordinates. This achieves a comprehensive and systematic integration of all key morphological parameters and electromagnetic feature parameters of microcracks, establishing a multi-dimensional quantitative conversion model. This allows for the simultaneous and accurate inversion of the complete three-dimensional morphology and spatial location of microcracks by measuring electromagnetic features, providing a core mathematical tool and judgment basis for subsequent high-precision quantitative detection.

[0224] Step 3.5: Based on the microcrack morphology-electromagnetic feature parameter mapping relationship matrix, normalize and encode the feature parameters in the electromagnetic wave response feature spectrum to establish the electromagnetic feature fingerprint database.

[0225] This embodiment normalizes and encodes the feature parameters in the electromagnetic wave response feature spectrum according to the microcrack morphology-electromagnetic feature parameter mapping relationship matrix, ultimately establishing a standardized electromagnetic feature fingerprint database. This fingerprint database unifies discrete, multi-dimensional electromagnetic feature data into a standardized system defined by the mapping relationship matrix, forming a "defect dictionary" that can be used for rapid comparison and intelligent identification, greatly improving the identification speed, consistency, and accuracy when subsequently detecting defects in unknown samples.

[0226] Step 4: In the accelerated aging test simulating the operating conditions of a dry-type transformer, the high thermal conductivity epoxy resin composite material sample was subjected to multi-stage electromagnetic wave scanning, and the electromagnetic wave scanning data was generated by encoding the data in time sequence.

[0227] Step 4.1: Fix the high thermal conductivity epoxy resin composite material sample in the aging test device according to the actual assembly position of the dry transformer winding.

[0228] This embodiment ensures that the mechanical, thermal, and electrical stresses experienced by the high thermal conductivity epoxy resin composite material sample during aging are consistent with the stress state in the actual product by fixing the sample in the aging test device according to the actual assembly orientation of the dry-type transformer winding. This allows the propagation path and final morphology of the microcracks generated in the accelerated aging test to highly reproduce the damage generated in actual operation, ensuring the authenticity and reliability of the subsequent electromagnetic wave scanning data and the aging evaluation results based on it.

[0229] Step 4.2: Set the environmental parameters for the aging test according to the actual operating conditions of the dry-type transformer.

[0230] In this embodiment, the environmental parameters include: a temperature cycling range of -40°C to +120°C, a relative humidity range of 20% to 95%, a mechanical vibration frequency of 10Hz to 200Hz and an amplitude of 0.5g to 5g, and a voltage level of 0.5 times to 1.5 times the rated voltage.

[0231] This embodiment constructs a harsh aging environment that comprehensively simulates the coupled effects of multiple factors such as temperature, humidity, mechanical vibration, and electrical stress by setting environmental parameters for the aging test based on the actual operating conditions of dry-type transformers. This multi-dimensional stress-accelerated test can effectively stimulate various aging failure modes that may occur in materials during actual operation, thereby obtaining highly representative aging data in a short period of time.

[0232] Step 4.3: Conduct accelerated aging tests according to the preset aging cycle.

[0233] In this embodiment, the aging cycle includes multiple consecutive aging stages, each of which includes a heating sub-stage, a heat preservation sub-stage, a cooling sub-stage, and a voltage loading sub-stage.

[0234] This embodiment employs accelerated aging tests according to a preset aging cycle, which includes multiple consecutive aging stages. Each aging stage comprises a heating sub-stage, a heat preservation sub-stage, a cooling sub-stage, and a voltage loading sub-stage, simulating the cyclical processes of start-up and shutdown, load changes, and diurnal temperature variations experienced by a dry-type transformer in actual operation. This periodic stress loading can more realistically accelerate the aging fatigue process of the material, promoting the more natural initiation and propagation of internal microcracks, thereby making the aging process closer to actual conditions and improving the accuracy of the life prediction model.

[0235] Step 4.4: After each aging stage, the high thermal conductivity epoxy resin composite material sample is scanned with electromagnetic waves using the multi-frequency electromagnetic wave scanning parameters to collect the original electromagnetic wave response signal.

[0236] Step 4.5: Perform signal preprocessing on the collected raw electromagnetic wave response signal to obtain the purified signal.

[0237] In this embodiment, the signal preprocessing includes wavelet thresholding noise reduction and matched filtering enhancement.

[0238] Step 4.6: Extract characteristic parameters, including characteristic frequency, characteristic amplitude, reflection coefficient and projection coefficient, from the purified signal to obtain the characteristic parameters of each stage.

[0239] This embodiment employs the multi-frequency electromagnetic wave scanning parameters after each aging stage to collect the original electromagnetic wave response signal. This signal is then preprocessed using the same methods as in the calibration stage to obtain a purified signal. Finally, feature parameters are extracted. This ensures that the data acquired throughout the aging test and the previously established electromagnetic fingerprint database originate from the same testing standards and data processing procedures. This consistency ensures effective comparison and fusion between static fingerprint database knowledge and dynamic aging monitoring data, providing high-quality, standardized input for subsequent deep learning models.

[0240] Step 4.7: According to the aging time series, the characteristic parameters of each stage are time-series encoded to generate electromagnetic wave scanning data.

[0241] This embodiment generates electromagnetic wave scanning data by encoding the characteristic parameters of each stage according to the aging time series. This integrates the discrete "snapshot" detection data at each aging time point into a time-series data stream that continuously reflects the trajectory of material performance degradation. This structured time-series data is key to driving the time-series analysis model, enabling it to learn and capture the trends and patterns of material aging. This provides a core data foundation for accurately assessing the degree of aging and scientifically predicting remaining lifespan.

[0242] Step 5: Use a deep learning model to detect and evaluate the electromagnetic feature fingerprint database and the electromagnetic wave scanning data to obtain quantitative detection results and quantitative evaluation results.

[0243] Step 5.1: Construct a deep learning model that includes a convolutional neural network branch and a long short-term memory network branch.

[0244] This embodiment creatively designs a hybrid neural network architecture capable of processing spatial static features and temporal dynamic features in parallel by constructing a deep learning model that includes convolutional neural network branches and long short-term memory network branches. It fully utilizes the advantages of convolutional neural networks in extracting spatial topological features from electromagnetic feature fingerprint databases, as well as the strengths of long short-term memory networks in analyzing temporal dependencies in electromagnetic wave scanning data, providing the optimal algorithmic foundation for subsequent comprehensive and accurate detection and evaluation.

[0245] Step 5.2: Input the electromagnetic feature fingerprint database into the convolutional neural network branch to extract spatial features and obtain static fingerprint features.

[0246] This embodiment obtains static fingerprint features by inputting the electromagnetic fingerprint database into the convolutional neural network branch for spatial feature extraction. This allows for the automatic learning and condensation of deep spatial patterns related to microcrack morphology contained in the fingerprint database. This step transforms prior calibration knowledge into a set of highly abstract and discriminative feature representations, providing stable and reliable benchmark reference information for subsequent fusion analysis.

[0247] Step 5.3: Input the electromagnetic wave scanning data into the long short-term memory network branch to extract the time-series features and obtain dynamic time-series features.

[0248] This embodiment extracts dynamic temporal features by inputting the electromagnetic wave scanning data into the long short-term memory network branch, effectively capturing and memorizing the evolution and long-term dependence of the electromagnetic characteristics of materials during aging. This step connects discrete time-point observations into a continuous degradation trajectory, providing a crucial dynamic information carrier for accurately assessing aging status and predicting remaining lifespan.

[0249] Step 5.4: Through a cross-modal attention fusion module, the static fingerprint features and dynamic temporal features are adaptively weighted and fused to obtain the fused feature vector.

[0250] This embodiment employs a cross-modal attention fusion module to adaptively weight and fuse static fingerprint features and dynamic temporal features to obtain a fused feature vector. This achieves deep complementarity and optimized integration of calibration prior knowledge and real-time monitoring data. The module automatically focuses on and strengthens feature dimensions more important to the current detection task, suppressing irrelevant or noisy information, thereby generating a more informative and discriminative unified feature representation, laying the foundation for the final high-precision output.

[0251] Step 5.5: Input the fused feature vector into a fully connected classification layer to classify the defect type and regress the spatial location, and output the type and location of the defects in the high thermal conductivity epoxy resin composite material sample.

[0252] This embodiment inputs the fused feature vector into a fully connected classification layer to classify defect types and regress spatial locations. By utilizing the information-rich fused features, it simultaneously completes the task of determining defect types and locations, realizing automated and parallel identification of the basic properties of internal defects in materials, and outputting the core qualitative information in the detection results.

[0253] Step 5.6: Input the fused feature vectors into a regression layer simultaneously to perform quantitative prediction of defect size and output the size of defects in the high thermal conductivity epoxy resin composite sample.

[0254] This embodiment inputs the fused feature vectors into a regression layer simultaneously to quantitatively predict the defect size. Based on the same fused feature, it achieves accurate inversion of the defect size in parallel, thus forming a complete static detection closed loop of "qualitative-localization-quantitative" microcrack detection.

[0255] Step 5.7: Input the dynamic time series features into a time series analysis layer to evaluate the aging degree of the high thermal conductivity epoxy resin composite sample and predict the remaining life, and output the predicted values ​​of the aging degree and remaining life of the high thermal conductivity epoxy resin composite sample.

[0256] This embodiment evaluates the aging degree of high thermal conductivity epoxy resin composite material samples and predicts their remaining life by inputting dynamic temporal features into a time series analysis layer. It specifically analyzes the dynamic process of material performance degradation. Based on the temporal evolution law extracted by the Long Short-Term Memory Network, it realizes the quantitative assessment of the aging degree of materials and the scientific inference of the remaining life, providing crucial assessment results.

[0257] Step 5.8: Integrate the outputs of the classification layer and the regression layer to obtain the quantitative detection results.

[0258] This embodiment integrates the outputs of the classification layer and the regression layer to obtain the quantitative detection results. The obtained defect type, location and size information are systematically summarized and correlated to form a complete and structured report on the internal defect state of the material, ensuring the comprehensiveness and usability of the detection results.

[0259] Step 5.9: Integrate the output of the time series analysis layer to obtain the quantitative evaluation results.

[0260] This embodiment integrates the output of the time series analysis layer to obtain the quantitative assessment results, standardizes the predicted information on material aging degree and remaining life, and finally completes the comprehensive assessment of the long-term operational reliability of the material, providing a direct and quantitative scientific basis for predictive maintenance decisions of the equipment.

[0261] In this embodiment, the quantitative detection results include the type, location, and size of defects in the high thermal conductivity epoxy resin composite material sample, and the quantitative evaluation results include the aging degree of the high thermal conductivity epoxy resin composite material sample and the predicted value of its remaining life.

[0262] This embodiment can also generate a health status report for the high thermal conductivity epoxy resin composite material based on the quantitative detection results and the quantitative assessment results. The health status report includes: A planar schematic diagram of a high thermal conductivity epoxy resin composite sample, including the type, location, and size of defects. Aging degree of high thermal conductivity epoxy resin composite material sample as a function of time; Predicted remaining life of high thermal conductivity epoxy resin composite material specimens.

[0263] This embodiment generates a health status report for high thermal conductivity epoxy resin composite materials based on the quantitative detection and evaluation results. This report systematically integrates key information such as a schematic diagram of the sample plane including defect types, locations, and sizes, aging degree over time curves, and predicted remaining life. It transforms complex detection data and evaluation conclusions into intuitive, multi-dimensional visualizations, achieving a leap from abstract data to concrete decision support. This provides equipment maintenance personnel with a clear global perspective on the current damage state and future performance evolution trends of the material, greatly improving the interpretability and practicality of the condition assessment results and providing the most direct and efficient basis for developing accurate predictive maintenance strategies.

[0264] Compared with the prior art, the beneficial effects of this embodiment are as follows: 1. Employing multi-frequency electromagnetic wave scanning with a bandwidth of 0.1~2.7THz, a resolution of 6GHz, and a dynamic range >60dB, and combining cluster analysis with a multi-level micro-defect information superposition algorithm to establish an electromagnetic feature fingerprint database, this technology overcomes the shortcomings of existing technologies, such as narrow bandwidth, low resolution, and difficulty in quantitatively characterizing microcracks. It achieves a precise defect detection resolution of <0.5mm for epoxy resin materials, and can accurately identify the type, location, and size of microcracks inside high thermal conductivity epoxy resin composite materials, significantly improving the accuracy and comprehensiveness of defect detection. 2. By simulating the actual operating conditions of dry-type transformers (temperature cycling, humidity, vibration, voltage coupling), accelerated aging tests are designed. Multi-stage electromagnetic wave scanning is performed on the samples and time-series encoded data is generated. This overcomes the shortcomings of existing technologies, such as the disconnect between aging tests and actual operating conditions and inaccurate life prediction. It can accurately track the material performance degradation trajectory, realize the scientific prediction of aging degree and remaining life, and provide a reliable basis for preventive maintenance. 3. A deep learning model incorporating convolutional neural networks and long short-term memory networks is constructed. By integrating static fingerprint features and dynamic temporal features through a cross-modal attention fusion module, the shortcomings of existing technologies, such as single feature dimensions and insufficient fusion, are overcome. This reduces the reliance on human experience and achieves automation and intelligence in detection and evaluation, ensuring the consistency and reliability of the results.

[0265] Further, in step S3, microscopic modeling, mesoscopic modeling, and macroscopic modeling are performed sequentially according to the material design parameters to obtain the formulation and structure of the high thermal conductivity epoxy resin. Specifically: Based on the material design parameters, the composite material component system is pre-designed to determine the basic formulation data, including the epoxy resin matrix type, the compound thermally conductive filler system, the interface modifier and its mass fraction range. Microscale molecular dynamics modeling was performed based on the basic formulation data to obtain microscale interface parameters including interfacial binding energy, interfacial thermal conductivity, matrix glass transition temperature, elastic modulus, and dielectric constant. Based on the microscopic interface parameters, a mesoscopic-scale phase-field method model is used to simulate the evolution of the material structure and obtain mesoscopic structural characteristic parameters including filler dispersion uniformity, interface layer thickness, aggregate size and distribution density. Based on the mesoscopic structural characteristic parameters, macroscopic-scale finite element modeling and multiphysics coupling analysis are performed to predict macroscopic performance parameters including thermal conductivity, bending strength and electrical strength. The macroscopic performance parameters are compared with the preset target performance, and the microscopic interface parameters, mesoscopic structure characteristic parameters and filler volume fraction are adjusted by orthogonal iteration method to perform parameter iterative optimization until the difference from the preset target performance is within the preset range, and the high thermal conductivity epoxy resin formulation and structure are output. The microscale molecular dynamics modeling includes analyzing the glass transition temperature of the matrix, the mesoscale phase field modeling includes controlling the evolution of material structure, and the macroscale finite element modeling includes multiphysics coupling analysis.

[0266] Further, in step S3, a high thermal conductivity epoxy resin composite material sample is prepared according to the stated formula and structure and subjected to terahertz detection. The sample evaluation result is obtained through deep learning algorithm analysis. Specifically: A high thermal conductivity epoxy resin composite material sample containing microcracks was prepared according to the aforementioned high thermal conductivity epoxy resin formulation and structure. Wideband terahertz time-domain spectroscopy was performed on the high thermal conductivity epoxy resin composite material sample. By optimizing the photoconductive antenna structure and signal processing algorithm, detection data with a bandwidth of 0.1-2.7THz, a resolution of 6GHz, and a dynamic range of >60dB were obtained. Based on the detection data, a defect fingerprint database containing electromagnetic wave characteristic parameters is constructed, including the reflection coefficient and the transmission coefficient. Based on the aging test data obtained from the defect fingerprint database and orthogonal test method, a mapping relationship between defect features and aging state is established using a deep learning algorithm; The high thermal conductivity epoxy resin composite material sample was tested according to the mapping relationship, and the sample evaluation results were obtained. The sample evaluation results include quantitative detection results and quantitative evaluation results; The quantitative detection results include the type, location, and size of defects in the high thermal conductivity epoxy resin composite material sample; The quantitative assessment results include the aging degree of the high thermal conductivity epoxy resin composite material sample and the predicted value of its remaining life.

[0267] Further, in step S3, the material design parameters are corrected based on the sample evaluation results to obtain a material performance dataset, and the dry-type transformer structure is designed by configuring the equivalent physical property parameters of the high thermal conductivity epoxy resin coil. Specifically: Based on the quantitative detection results in the sample evaluation results, the filler selection and interface modifier parameters in the high thermal conductivity epoxy resin formulation are optimized by feedback to obtain the corrected material design parameters. Based on the revised material design parameters and the aging degree of the high thermal conductivity epoxy resin composite material specimens in the quantitative evaluation results, a material property dataset including thermal conductivity, electrical strength, and flexural strength is established. Based on the aforementioned material property dataset, the equivalent physical property parameters of a high thermal conductivity epoxy resin coil are configured by studying the thermal conductivity characteristics under different coil winding methods and air passage arrangement methods. Based on the equivalent physical property parameters and the results of the study on the thermal conductivity characteristics of the core and coil, the overall structure of the dry-type transformer is designed, resulting in a dry-type transformer structure that includes the core structure, coil arrangement, and air duct layout.

[0268] Further, in step S3, based on the material property dataset, by studying the thermal conductivity characteristics under different coil winding methods and air passage arrangements, the equivalent physical property parameters of the high thermal conductivity epoxy resin coil are configured, specifically: Based on the thermal conductivity in the material property dataset, a porous medium free fluid coupled flow heat transfer model is established, which includes wires, insulating films, high thermal conductivity epoxy resin composite materials, and cooling air channels. Based on the porous medium free fluid coupled flow heat transfer model, and considering the gas flow characteristics under different coil winding methods (layered winding, segmented winding), and different air passage spacing and number arrangement, the gas permeability and heat dispersion coefficient are determined. Based on the gas permeability and heat dispersion coefficient, the thermal conductivity parameters of the high thermal conductivity epoxy resin coil are determined by solving the non-thermal equilibrium flow heat transfer equation and considering the coil temperature distribution characteristics under different winding and arrangement methods. Based on the aforementioned thermal conductivity parameters, and configured according to constraints including coil temperature rise limits, main insulation strength, and inter-turn insulation strength, the equivalent physical property parameters of the high thermal conductivity epoxy resin coil are obtained.

[0269] Further, in step S3, based on the dry-type transformer structure and the harmonics and power output fluctuations in the renewable energy access scenario, a multi-physics coupled simulation model is constructed and its performance is verified to obtain the performance evaluation results of the dry-type transformer. Specifically: Based on the dry-type transformer structure and the equivalent physical property parameters of the high thermal conductivity epoxy resin coil, a multi-physics field coupling simulation model including electric field, magnetic field, temperature field and stress field is constructed. Based on the harmonic frequency, harmonic content, and power output fluctuation amplitude parameters of the renewable energy access scenario, the boundary conditions and excitation sources of the multiphysics coupling simulation model are set. Based on the boundary conditions and excitation sources of the multiphysics coupling simulation model, the multiphysics coupling simulation model is run and electric field analysis is performed to obtain the electric field intensity distribution and inter-turn voltage distribution. Based on the boundary conditions and excitation source of the multiphysics coupling simulation model, the multiphysics coupling simulation model is run and magnetic field analysis is performed to obtain the magnetic flux density distribution and core loss. Based on the boundary conditions and excitation sources of the multiphysics coupled simulation model, the multiphysics coupled simulation model is run and the temperature field is analyzed to obtain the temperature distribution and hot spot temperature rise. Based on the boundary conditions and excitation sources of the multiphysics coupled simulation model, the multiphysics coupled simulation model is run and stress field analysis is performed to obtain the mechanical stress distribution and deformation displacement. Based on the electric field intensity distribution, inter-turn voltage distribution, magnetic flux density distribution, temperature distribution, hot spot temperature rise, mechanical stress distribution, and deformation displacement, the performance evaluation results of the dry-type transformer are verified.

[0270] Further, in step S3, based on the electric field intensity distribution, inter-turn voltage distribution, magnetic flux density distribution, temperature distribution, hot spot temperature rise, mechanical stress distribution, and deformation displacement, the performance evaluation results of the dry-type transformer are verified, specifically: Based on the electric field strength distribution, check whether the maximum electric field strength is lower than the breakdown field strength threshold of the high thermal conductivity epoxy resin composite material, and obtain the electric field strength check result; Based on the inter-turn voltage distribution, check whether the inter-turn potential difference meets the insulation coordination requirements, and obtain the inter-turn insulation check result; Based on the temperature distribution and hot spot temperature rise, check whether the temperature rise of the highest temperature point is lower than the preset temperature rise limit to obtain the temperature rise performance verification result. Based on the mechanical stress distribution and deformation displacement, check whether the maximum stress value is lower than the yield strength of the high thermal conductivity epoxy resin composite material and whether the deformation is within the allowable range, and obtain the mechanical stability check result; Based on the electric field strength verification results, inter-turn insulation verification results, temperature rise performance verification results, and mechanical stability verification results, a comprehensive evaluation is conducted to obtain the performance assessment results of the dry-type transformer: If all verification results pass, the dry-type transformer is deemed to meet the operating requirements, and the performance evaluation result is deemed qualified. If any verification result fails, the dry-type transformer is deemed not to meet the operating requirements, and the performance evaluation result is deemed unqualified. Based on the failed verification results, the specific substandard performance indicators and their distribution areas in the dry-type transformer structure are identified. The performance indicators include electric field strength, inter-turn insulation, hot spot temperature rise, or mechanical stress.

[0271] Further, in step S3, based on the performance evaluation results, with the optimization objectives of maximizing equipment capacity and minimizing volume, and with electric field strength, temperature rise limit, and mechanical stress as constraints, a multi-objective collaborative optimization is performed using the Monte Carlo iterative method and a global optimization algorithm to obtain the optimized global design parameters for the dry-type transformer. Specifically: If the performance evaluation result is qualified, the current dry-type transformer structural parameters will be used as the optimized global design parameters for the dry-type transformer. If the performance evaluation result is unqualified, the structural parameters of the dry-type transformer to be optimized are determined based on the specific unqualified performance indicators and their distribution areas in the dry-type transformer structure. The structural parameters of the dry-type transformer to be optimized include the epoxy resin content of the core, the coil winding method, the air passage layout parameters, the wire diameter and the wire end distance. Based on the structural parameters of the dry-type transformer to be optimized, a multi-objective optimization function is established with the optimization objectives of maximizing equipment capacity and minimizing volume, and with constraints of electric field strength not exceeding the standard, temperature rise not exceeding the limit, and mechanical stress less than the material tolerance value. Based on the multi-objective optimization function, the Monte Carlo iteration method is used to randomly sample within the feasible region of the dry-type transformer structural parameters to be optimized, generating multiple candidate dry-type transformer structural parameter design schemes. According to the global optimization algorithm, the optimization calculation is performed on the multiple candidate dry-type transformer structural parameter design schemes to obtain the candidate dry-type transformer structural parameter design schemes that satisfy all constraints and make the optimization objective optimal, which are used as the optimized global design parameters of the dry-type transformer.

[0272] Further, in step S3, a prototype is manufactured according to the optimized global design parameters of the dry-type transformer, and a power frequency-harmonic composite current temperature rise test and a high-frequency oscillation pulse voltage inter-turn insulation test are performed sequentially to obtain a high thermal conductivity epoxy resin dry-type transformer prototype that has been tested and verified. Specifically: Based on the optimized global design parameters of the dry-type transformer, a high thermal conductivity epoxy resin dry-type transformer prototype was manufactured by using high thermal conductivity epoxy resin composite material for coil winding, core assembly and vacuum casting and curing. According to the high thermal conductivity epoxy resin dry molding machine, a power frequency-harmonic composite current temperature rise test was conducted, including: A high-voltage synthetic circuit is constructed by using the harmonic compensation method, which superimposes the 3rd, 5th, and 7th characteristic harmonic currents onto the power frequency current. Pre-embedded thermocouples and infrared thermal imagers were used to measure the temperature rise distribution of the windings, and resistance strain gauges and Raman distributed fiber optic sensors were used to monitor changes in thermal stress. Run continuously until the temperature rise stabilizes, record temperature and stress data, establish the harmonic-temperature rise correspondence and thermal stress distribution model, and obtain the temperature rise and stress assessment results; A high-frequency oscillating pulse voltage inter-turn insulation test was performed using the aforementioned high thermal conductivity epoxy resin dry-type molding machine, including: An oscillating pulse voltage is generated using a high-voltage synthesis circuit with a trigger gap; The partial discharge initiation voltage was measured by boosting the voltage at a rate of 1kV / s, and lightning impulse and switching impulse voltages were applied. By monitoring the discharge signal with a partial discharge detector, the coil was disassembled after the test to observe the carbonization points, an insulation failure early warning threshold was determined, and the inter-turn insulation assessment results were obtained. Based on the temperature rise and stress assessment results and the inter-turn insulation assessment results, a comprehensive determination is made as to whether the high thermal conductivity epoxy resin dry converter prototype passes the test verification: If the test verification is passed, the high thermal conductivity epoxy resin dry deformation sampler shall be regarded as the high thermal conductivity epoxy resin dry deformation sampler verified by the test. If the test verification fails, the specific failure mode of the high thermal conductivity epoxy resin dry-type transformer prototype will be located based on the failure to meet the temperature rise and stress assessment results and / or the failure to meet the inter-turn insulation assessment results, and the global design parameters of the dry-type transformer will be optimized accordingly.

[0273] For example: This embodiment introduces a dry-varying design method and verification method for high thermal conductivity epoxy resin, including: Step 1: Based on the material design parameters, perform microscopic modeling, mesoscopic modeling, and macroscopic modeling in sequence to obtain the formulation and structure of the high thermal conductivity epoxy resin.

[0274] By sequentially performing multi-scale modeling processes—microscale molecular dynamics modeling, mesoscale phase-field modeling, and macroscale finite element modeling—a complete predictive model from molecular interface parameters to macroscopic properties was established. This enabled the scientific design of high thermal conductivity epoxy resin formulations and structures, solving the problems of long development cycles and high costs associated with traditional trial-and-error methods.

[0275] Step 2: Prepare high thermal conductivity epoxy resin composite material samples according to the formula and structure, and perform terahertz detection. Analyze the samples using a deep learning algorithm to obtain the evaluation results.

[0276] By preparing high thermal conductivity epoxy resin composite material samples according to the formula and structure and performing terahertz detection, and combining deep learning algorithm analysis to obtain sample evaluation results, the accurate identification of microcrack defects inside the material and the quantitative evaluation of aging state were realized, providing reliable data support for material performance optimization.

[0277] Step 3: Based on the evaluation results of the test samples, the material design parameters are corrected to obtain the material performance dataset, and the dry-type transformer structure is designed by configuring the equivalent physical property parameters of the high thermal conductivity epoxy resin coil.

[0278] By modifying the material design parameters based on the sample evaluation results, a material performance dataset was obtained. The equivalent physical property parameters of the high thermal conductivity epoxy resin coil were configured to design the dry-type transformer structure, establishing a direct correlation between material performance and equipment design, and realizing integrated material-structure optimization design.

[0279] Step 4: Based on the dry-type transformer structure, and considering the harmonics and power output fluctuations in the renewable energy access scenario, construct a multi-physics field coupled simulation model and perform performance verification to obtain the performance evaluation results of the dry-type transformer.

[0280] By constructing a multi-physics field coupled simulation model based on the dry-type transformer structure and the harmonics and power output fluctuations in the renewable energy access scenario, and performing performance verification, the performance evaluation results of the dry-type transformer are obtained. This achieves accurate prediction of equipment performance under complex working conditions and solves the problem of inaccurate traditional single-field verification.

[0281] Step 5: Based on the performance evaluation results, with the optimization objectives of maximizing equipment capacity and minimizing volume, and with electric field strength, temperature rise limit and mechanical stress as constraints, the Monte Carlo iterative method and global optimization algorithm are used to perform multi-objective collaborative optimization to obtain the optimized global design parameters of the dry-type transformer.

[0282] Based on the performance evaluation results, with the optimization objectives of maximizing equipment capacity and minimizing volume, and with electric field strength, temperature rise limit and mechanical stress as constraints, the Monte Carlo iterative method and global optimization algorithm are used for multi-objective collaborative optimization, which realizes the optimization of global design parameters of dry transformers and solves the problem that traditional methods are difficult to achieve global optimization.

[0283] Step Six: Based on the optimized global design parameters of the dry-type transformer, a prototype is manufactured, and the power frequency-harmonic composite current temperature rise test and the high-frequency oscillation pulse voltage inter-turn insulation test are carried out in sequence to obtain a high thermal conductivity epoxy resin dry-type transformer prototype that has been tested and verified.

[0284] By manufacturing a prototype based on the optimized global design parameters of the dry-type transformer, and conducting sequential power frequency-harmonic composite current temperature rise test and high frequency oscillation pulse voltage inter-turn insulation test, the prototype was fully verified under real working conditions, ensuring the reliability of the high thermal conductivity epoxy resin dry-type transformer prototype.

[0285] The implementation steps of a dry-varying design method and verification method for high thermal conductivity epoxy resin include: Step 1: Based on the material design parameters, perform microscopic modeling, mesoscopic modeling, and macroscopic modeling in sequence to obtain the formulation and structure of the high thermal conductivity epoxy resin.

[0286] In this embodiment, the material design parameters include epoxy resin matrix parameters, compound thermally conductive filler parameters, and interface modifier parameters.

[0287] Step 1.1: Based on the material design parameters, pre-design the composite material component system and determine the basic formulation data, including the epoxy resin matrix type, the compound thermally conductive filler system, the interface modifier and its mass fraction range.

[0288] This embodiment pre-designs the composite material component system based on material design parameters, determining the basic formulation data including epoxy resin matrix type, compound thermally conductive filler system, interface modifier and its mass fraction range, realizing the systematic design of material formulation, and providing accurate input parameters for subsequent multi-scale modeling.

[0289] Step 1.2: Perform microscale molecular dynamics modeling based on the basic formulation data, and analyze the microscale interface parameters including interfacial binding energy, interfacial thermal conductivity, matrix glass transition temperature, elastic modulus and dielectric constant.

[0290] This embodiment uses microscale molecular dynamics modeling based on basic formulation data to analyze and obtain microscale interface parameters including interfacial binding energy, interfacial thermal conductivity, matrix glass transition temperature, elastic modulus, and dielectric constant. This reveals the material interface properties at the molecular level and provides key input parameters for mesoscale modeling.

[0291] Step 1.3: Based on the microscopic interface parameters, perform mesoscopic-scale phase-field modeling to simulate the evolution of the material structure and obtain mesoscopic structural characteristic parameters including filler dispersion uniformity, interface layer thickness, agglomerate size and distribution density.

[0292] This embodiment uses mesoscale phase-field modeling based on microscopic interface parameters to simulate the evolution of material structure, obtaining mesoscopic structural characteristic parameters including filler dispersion uniformity, interface layer thickness, agglomerate size and distribution density, thus achieving accurate prediction of microstructure formation during material preparation.

[0293] Step 1.4: Perform macroscopic finite element modeling and multiphysics coupling analysis based on the mesoscopic structural characteristic parameters to predict macroscopic performance parameters including thermal conductivity, bending strength and electrical strength.

[0294] This embodiment predicts macroscopic performance parameters, including thermal conductivity, bending strength, and electrical strength, by performing macroscopic-scale finite element modeling and multiphysics coupling analysis based on mesoscopic structural characteristic parameters, thus establishing an accurate correlation between microstructure and macroscopic performance.

[0295] Step 1.5: Compare the macroscopic performance parameters with the preset target performance, and adjust the microscopic interface parameters, mesoscopic structure characteristic parameters and filler volume fraction through orthogonal iteration method to perform parameter iterative optimization until the difference from the preset target performance is within the preset range, and output the high thermal conductivity epoxy resin formulation and structure.

[0296] This embodiment compares macroscopic performance parameters with preset target performance and uses an orthogonal iterative method to adjust microscopic interface parameters, mesoscopic structural characteristic parameters, and filler volume fraction, thereby achieving precise control of material properties and ensuring that the final formulation and structure meet design requirements.

[0297] Step 1.6: Perform microscale molecular dynamics modeling and interface parameter analysis based on the basic formulation data to obtain microscale interface parameters including interfacial binding energy, interfacial thermal conductivity, matrix glass transition temperature, elastic modulus, and dielectric constant.

[0298] This embodiment obtains microscopic interface parameters, including interfacial binding energy, interfacial thermal conductivity, matrix glass transition temperature, elastic modulus, and dielectric constant, by performing microscale molecular dynamics modeling and interface parameter analysis based on basic formulation data. This provides reliable underlying parameter support for multiscale modeling systems.

[0299] Step 1.7: Based on the microscopic interface parameters, perform mesoscopic-scale phase-field modeling and structural evolution control to obtain mesoscopic structural characteristic parameters including filler dispersion uniformity, interface layer thickness, aggregate size and distribution density.

[0300] This embodiment achieves controllable design of the material's microstructure by using mesoscale phase-field modeling and structural evolution control based on micro-interface parameters to obtain mesoscopic structural characteristic parameters, including filler dispersion uniformity, interface layer thickness, aggregate size and distribution density.

[0301] Step 1.8: Perform macroscopic finite element modeling and performance prediction based on the mesoscopic structural characteristic parameters to obtain macroscopic performance parameters including thermal conductivity, bending strength and electrical strength.

[0302] This embodiment establishes a complete material performance prediction system by performing macroscopic finite element modeling and performance prediction based on mesoscopic structural characteristic parameters, obtaining macroscopic performance parameters including thermal conductivity, bending strength, and electrical strength.

[0303] Step 1.9: Based on the comparison results between the macroscopic performance parameters and the preset target performance, perform parameter iterative optimization and performance control. Through orthogonal iteration, adjust the microscopic interface parameters, mesoscopic structural characteristic parameters and macroscopic performance parameters to obtain the formulation and structure of high thermal conductivity epoxy resin that meets the preset target performance.

[0304] This embodiment optimizes and controls parameters iteratively based on the comparison results between macroscopic performance parameters and preset target performance. It uses an orthogonal iterative method to adjust microscopic interface parameters, mesoscopic structural characteristic parameters, and macroscopic performance parameters, forming a complete material design closed loop to ensure that a high thermal conductivity epoxy resin formulation and structure that meets the preset target performance is obtained.

[0305] Step 2: Prepare high thermal conductivity epoxy resin composite material samples according to the formula and structure and perform terahertz detection. Analyze the samples using a deep learning algorithm to obtain the sample evaluation results.

[0306] In this embodiment, the sample evaluation results include the aging degree of the high thermal conductivity epoxy resin composite sample and the predicted value of its remaining life.

[0307] Step 2.1: Prepare a high thermal conductivity epoxy resin composite material sample containing microcracks according to the high thermal conductivity epoxy resin formulation and structure.

[0308] This embodiment prepares a high thermal conductivity epoxy resin composite material sample containing microcracks according to the high thermal conductivity epoxy resin formulation and structure, providing a standardized test sample for subsequent testing and analysis, and ensuring the reliability and comparability of the test results.

[0309] Step 2.2: Broadband terahertz time-domain spectroscopy was performed on the high thermal conductivity epoxy resin composite material sample. By optimizing the photoconductive antenna structure and signal processing algorithm, detection data with a bandwidth of 0.1-2.7THz, a resolution of 6GHz, and a dynamic range of >60dB were obtained.

[0310] This embodiment achieves high-precision detection of internal defects in materials by performing broadband terahertz time-domain spectroscopy on the high thermal conductivity epoxy resin composite material sample, and by optimizing the photoconductive antenna structure and signal processing algorithm, obtaining detection data with a bandwidth of 0.1-2.7THz, a resolution of 6GHz, and a dynamic range of >60dB.

[0311] Step 2.3: Construct a defect fingerprint database containing electromagnetic wave characteristic parameters based on the detection data. The electromagnetic wave characteristic parameters include reflection coefficient and transmission coefficient.

[0312] This embodiment constructs a defect fingerprint database containing electromagnetic wave characteristic parameters based on the detection data. These electromagnetic wave characteristic parameters include reflection coefficient and transmission coefficient, thus establishing a complete defect feature database and providing a data foundation for subsequent defect identification.

[0313] Step 2.4: Based on the defect fingerprint database and the aging test data obtained by the orthogonal experimental method, establish the mapping relationship between defect features and aging state through deep learning algorithm.

[0314] This embodiment establishes a mapping relationship between defect features and aging state using deep learning algorithms based on the aging test data obtained from the defect fingerprint database and orthogonal experimental method, thereby realizing intelligent assessment of the material aging state.

[0315] Step 2.5: Test the high thermal conductivity epoxy resin composite material sample according to the mapping relationship to obtain the sample evaluation results.

[0316] This embodiment completes the entire analysis process from test data to material condition assessment by testing a high thermal conductivity epoxy resin composite material sample according to the mapping relationship and obtaining the sample evaluation results.

[0317] In this embodiment, the sample evaluation results include quantitative detection results and quantitative evaluation results; the quantitative detection results include the type, location, and size of defects in the high thermal conductivity epoxy resin composite material sample; the quantitative evaluation results include the aging degree of the high thermal conductivity epoxy resin composite material sample and the predicted value of its remaining life.

[0318] Step 3: Based on the evaluation results of the sample, correct the material design parameters to obtain the material performance dataset, and design the dry-type transformer structure by configuring the equivalent physical property parameters of the high thermal conductivity epoxy resin coil.

[0319] In this embodiment, the equivalent physical property parameters include the winding method, air passage arrangement, and core structure.

[0320] Step 3.1: Based on the quantitative detection results in the sample evaluation results, the filler selection and interface modifier parameters in the high thermal conductivity epoxy resin formulation are optimized to obtain the corrected material design parameters.

[0321] This embodiment optimizes the filler selection and interface modifier parameters in the high thermal conductivity epoxy resin formulation based on the quantitative detection results in the sample evaluation results, obtains the corrected material design parameters, establishes a direct correlation between material testing results and formulation optimization, and achieves continuous improvement of material performance.

[0322] Step 3.2: Based on the revised material design parameters and the aging degree of the high thermal conductivity epoxy resin composite material sample in the quantitative evaluation results, establish a material property dataset including thermal conductivity, electrical strength and flexural strength.

[0323] This embodiment establishes a material performance dataset including thermal conductivity, electrical strength, and flexural strength based on the revised material design parameters and the aging degree of the high thermal conductivity epoxy resin composite material sample in the quantitative evaluation results, forming a complete material performance database that provides reliable data support for equipment design.

[0324] Step 3.3: Based on the material property dataset, the equivalent physical property parameters of the high thermal conductivity epoxy resin coil are configured by studying the thermal conductivity characteristics under different coil winding methods and air passage arrangements.

[0325] Step 3.3.1: Based on the thermal conductivity in the material property dataset, establish a porous medium free fluid coupled flow heat transfer model that includes wires, insulating films, high thermal conductivity epoxy resin composite materials, and cooling air channels.

[0326] This embodiment establishes a porous medium free fluid coupled flow heat transfer model, which includes wires, insulating films, high thermal conductivity epoxy resin composite materials, and cooling air channels, based on the thermal conductivity of the material property dataset, thereby achieving accurate simulation of the coil heat dissipation characteristics.

[0327] Step 3.3.2: Based on the porous medium free fluid coupled flow heat transfer model, and considering the gas flow characteristics under different coil winding methods (layered winding, segmented winding) and different air channel spacing and number arrangement, determine the gas permeability and heat dispersion coefficient.

[0328] This embodiment determines the gas permeability and heat dissipation coefficient based on the porous medium free fluid coupled flow heat transfer model, and on the gas flow characteristics under different coil winding methods such as layered winding and segmented winding, as well as different air channel spacing and air channel number arrangement, thereby quantifying the influence of different structural parameters on heat dissipation performance.

[0329] Step 3.3.3: Based on the gas permeability and heat dispersion coefficient, the thermal conductivity parameters of the high thermal conductivity epoxy resin coil are determined by solving the non-thermal equilibrium flow heat transfer equation and considering the coil temperature distribution characteristics under different winding and arrangement methods.

[0330] This embodiment determines the thermal conductivity parameters of a high thermal conductivity epoxy resin coil by solving the non-thermal equilibrium flow heat transfer equation based on the gas permeability and heat dispersion coefficient, and by considering the coil temperature distribution characteristics under different winding and arrangement methods. This enables accurate prediction of the coil's thermal performance.

[0331] Step 3.3.4: Based on the thermal conductivity parameters, and considering constraints including coil temperature rise limit, main insulation strength and inter-turn insulation strength, the equivalent physical property parameters of the high thermal conductivity epoxy resin coil are obtained.

[0332] This embodiment obtains the equivalent physical property parameters of the high thermal conductivity epoxy resin coil by configuring the parameters based on the thermal conductivity characteristics, including coil temperature rise limits, main insulation strength and inter-turn insulation strength constraints, thus ensuring that the coil design meets multiple requirements.

[0333] Step 3.4: Based on the equivalent physical property parameters and the results of the thermal conductivity study of the core and coil, the overall structure of the dry-type transformer is designed to obtain a dry-type transformer structure including the core structure, coil arrangement and air passage layout.

[0334] This embodiment, based on the equivalent physical property parameters and the research results on the thermal conductivity characteristics of the core and coil, conducts the overall structural design of the dry-type transformer, resulting in a dry-type transformer structure that includes the core structure, coil arrangement, and air duct layout, thus completing the complete design process from material properties to equipment structure.

[0335] Step 4: Based on the dry-type transformer structure, and considering the harmonics and power output fluctuations in the renewable energy access scenario, construct a multi-physics field coupled simulation model and perform performance verification to obtain the performance evaluation results of the dry-type transformer.

[0336] Step 4.1: Based on the dry-type transformer structure and the equivalent physical property parameters of the high thermal conductivity epoxy resin coil, construct a multi-physics field coupling simulation model including electric field, magnetic field, temperature field and stress field.

[0337] This embodiment constructs a multi-physics field coupled simulation model, including electric field, magnetic field, temperature field and stress field, based on the equivalent physical property parameters of the dry-type transformer structure and the high thermal conductivity epoxy resin coil, thereby realizing the all-round performance simulation of the dry-type transformer under complex working conditions.

[0338] Step 4.2: Based on the harmonic frequency, harmonic content, and power output fluctuation amplitude parameters of the renewable energy access scenario, set the boundary conditions and excitation sources of the multiphysics coupling simulation model.

[0339] This embodiment sets the boundary conditions and excitation sources of the multiphysics coupling simulation model based on the harmonic frequency, harmonic content, and power output fluctuation amplitude parameters of the renewable energy access scenario, thus ensuring the consistency between the simulation model and the actual operating conditions.

[0340] Step 4.3: Based on the boundary conditions and excitation sources of the multiphysics coupling simulation model, run the multiphysics coupling simulation model and perform electric field analysis to obtain the electric field intensity distribution and inter-turn voltage distribution.

[0341] This embodiment uses the boundary conditions and excitation source of the physical field coupling simulation model to run the multi-physics field coupling simulation model and perform electric field analysis to solve for the electric field intensity distribution and inter-turn voltage distribution, thus accurately evaluating the insulation performance of the dry-type transformer.

[0342] Step 4.4: Based on the boundary conditions and excitation source of the multiphysics coupling simulation model, run the multiphysics coupling simulation model and perform magnetic field analysis to obtain the magnetic flux density distribution and core loss.

[0343] This embodiment uses the boundary conditions and excitation source of the physical field coupling simulation model to run the multi-physics field coupling simulation model and perform magnetic field analysis to solve for the magnetic flux density distribution and core loss, providing an important basis for the magnetic circuit design of dry-type transformers.

[0344] Step 4.5: Based on the boundary conditions and excitation sources of the multiphysics coupling simulation model, run the multiphysics coupling simulation model and perform temperature field analysis to obtain the temperature distribution and hot spot temperature rise.

[0345] This embodiment uses the boundary conditions and excitation source of the physical field coupling simulation model to run the multiphysics field coupling simulation model and perform temperature field analysis to solve for the temperature distribution and hot spot temperature rise, thus realizing the accurate prediction of the thermal performance of dry-type transformers.

[0346] Step 4.6: Based on the boundary conditions and excitation sources of the multiphysics coupling simulation model, run the multiphysics coupling simulation model and perform stress field analysis to obtain the mechanical stress distribution and deformation displacement.

[0347] This embodiment uses the boundary conditions and excitation sources of the multiphysics coupling simulation model to run the model and perform stress field analysis, thereby obtaining the mechanical stress distribution and deformation displacement, providing reliable data for the structural strength design of dry-type transformers.

[0348] Step 4.7: Based on the electric field intensity distribution, inter-turn voltage distribution, magnetic flux density distribution, temperature distribution, hot spot temperature rise, mechanical stress distribution, and deformation displacement, the performance evaluation results of the dry-type transformer are verified.

[0349] Step 4.7.1: Based on the electric field strength distribution, check whether the maximum electric field strength is lower than the breakdown field strength threshold of the high thermal conductivity epoxy resin composite material, and obtain the electric field strength verification result.

[0350] This embodiment verifies whether the maximum electric field strength is lower than the breakdown field strength threshold of the high thermal conductivity epoxy resin composite material based on the electric field strength distribution, thereby obtaining the electric field strength verification result and ensuring the insulation reliability of the dry-type transformer.

[0351] Step 4.7.2: Based on the inter-turn voltage distribution, check whether the inter-turn potential difference meets the insulation coordination requirements, and obtain the inter-turn insulation check result.

[0352] This embodiment verifies whether the inter-turn potential difference meets the insulation coordination requirements based on the inter-turn voltage distribution, thereby obtaining the inter-turn insulation verification result and effectively preventing the occurrence of inter-turn insulation faults.

[0353] Step 4.7.3: Based on the temperature distribution and hot spot temperature rise, check whether the temperature rise of the highest temperature point is lower than the preset temperature rise limit, and obtain the temperature rise performance verification result.

[0354] This embodiment verifies whether the temperature rise at the highest temperature point is lower than the preset temperature rise limit based on the temperature distribution and hot spot temperature rise, thus obtaining the temperature rise performance verification result and ensuring the thermal stability of the dry-type transformer.

[0355] Step 4.7.4: Based on the mechanical stress distribution and deformation displacement, check whether the maximum stress value is lower than the yield strength of the high thermal conductivity epoxy resin composite material and whether the deformation is within the allowable range, and obtain the mechanical stability check result.

[0356] This embodiment verifies whether the maximum stress value is lower than the yield strength of the high thermal conductivity epoxy resin composite material and whether the deformation is within the allowable range based on the mechanical stress distribution and deformation displacement, thus obtaining the mechanical stability verification result and ensuring the structural integrity of the dry-type transformer.

[0357] Step 4.7.5: Based on the electric field strength verification results, inter-turn insulation verification results, temperature rise performance verification results, and mechanical stability verification results, a comprehensive evaluation is performed to obtain the performance assessment results of the dry-type transformer: If all verification results pass, the dry-type transformer is deemed to meet the operating requirements, and the performance evaluation result is deemed qualified. If any verification result fails, the dry-type transformer is deemed not to meet the operating requirements, and the performance evaluation result is deemed unqualified. Based on the failed verification results, the specific substandard performance indicators and their distribution areas in the dry-type transformer structure are identified.

[0358] This embodiment obtains the performance evaluation results of the dry-type transformer by comprehensively evaluating the electric field strength verification results, inter-turn insulation verification results, temperature rise performance verification results, and mechanical stability verification results. It establishes a complete performance evaluation system and provides a clear direction for design optimization.

[0359] In this embodiment, the performance indicators include electric field strength, inter-turn insulation, hot spot temperature rise or mechanical stress, and the performance evaluation results include hot spot temperature rise, stress distribution and inter-turn voltage performance.

[0360] Step 5: Based on the performance evaluation results, with the optimization objectives of maximizing equipment capacity and minimizing volume, and with electric field strength, temperature rise limit and mechanical stress as constraints, the Monte Carlo iterative method and global optimization algorithm are used to perform multi-objective collaborative optimization to obtain the optimized global design parameters of the dry-type transformer.

[0361] Step 5.1: If the performance evaluation result is qualified, the current dry-type transformer structural parameters are used as the optimized global design parameters of the dry-type transformer.

[0362] This embodiment uses the current dry-type transformer structural parameters as the optimized global design parameters if the performance evaluation result is qualified, thus avoiding unnecessary optimization processes and improving design efficiency.

[0363] Step 5.2: If the performance evaluation result is unqualified, then determine the structural parameters of the dry-type transformer to be optimized based on the specific unqualified performance indicators and their distribution areas in the dry-type transformer structure. The structural parameters of the dry-type transformer to be optimized include the epoxy resin content of the core, the coil winding method, the air passage layout parameters, the wire diameter and the distance between the wire ends.

[0364] In this embodiment, if the performance evaluation result is unqualified, the structural parameters of the dry-type transformer to be optimized are determined based on the specific substandard performance indicators and their distribution areas in the dry-type transformer structure. The structural parameters of the dry-type transformer to be optimized include the epoxy resin content of the core, the coil winding method, the air passage layout parameters, the wire diameter and the wire end distance, thus achieving targeted structural optimization.

[0365] Step 5.3: Based on the structural parameters of the dry-type transformer to be optimized, establish a multi-objective optimization function with the optimization objectives of maximizing equipment capacity and minimizing volume, and with constraints of electric field strength not exceeding the standard, temperature rise not exceeding the limit, and mechanical stress less than the material tolerance value.

[0366] This embodiment establishes a multi-objective optimization function based on the structural parameters of the dry-type transformer to be optimized. The optimization objectives are to maximize equipment capacity and minimize volume, while the constraints are that the electric field strength does not exceed the standard, the temperature rise does not exceed the limit, and the mechanical stress is less than the material tolerance value. A complete optimization mathematical model is thus constructed.

[0367] Step 5.4: Based on the multi-objective optimization function, the Monte Carlo iteration method is used to randomly sample within the feasible region of the dry-type transformer structural parameters to be optimized, generating multiple candidate dry-type transformer structural parameter design schemes.

[0368] This embodiment generates multiple candidate dry-type transformer structural parameter design schemes by randomly sampling within the feasible region of the dry-type transformer structural parameters to be optimized according to the multi-objective optimization function using the Monte Carlo iteration method, thereby achieving a full exploration of the design space.

[0369] Step 5.5: Based on the global optimization algorithm, perform optimization calculations on the multiple candidate dry-type transformer structural parameter design schemes, and solve for the candidate dry-type transformer structural parameter design schemes that satisfy all constraints and make the optimization objective optimal, which are then used as the optimized global design parameters for the dry-type transformer.

[0370] In this embodiment, the global optimization algorithm is used to perform optimization calculations on the multiple candidate dry-type transformer structural parameter design schemes. The solution is obtained to obtain the candidate dry-type transformer structural parameter design scheme that satisfies all constraints and makes the optimization objective optimal. This is used as the optimized global design parameters for the dry-type transformer, thus obtaining the globally optimal design scheme.

[0371] Step 6: Based on the optimized global design parameters of the dry-type transformer, a prototype is manufactured, and the power frequency-harmonic composite current temperature rise test and the high-frequency oscillation pulse voltage inter-turn insulation test are carried out in sequence to obtain a high thermal conductivity epoxy resin dry-type transformer prototype that has been verified by the test.

[0372] Step 6.1: Based on the optimized global design parameters of the dry converter, a high thermal conductivity epoxy resin composite material is used for coil winding, core assembly and vacuum casting curing to manufacture a high thermal conductivity epoxy resin dry converter prototype.

[0373] This embodiment uses high thermal conductivity epoxy resin composite material for coil winding, core assembly, and vacuum casting and curing based on the optimized global design parameters of the dry-type transformer, to manufacture a high thermal conductivity epoxy resin dry-type transformer prototype. This achieves accurate conversion of design parameters into a physical prototype and ensures the quality of prototype manufacturing.

[0374] Step 6.2: Using the high thermal conductivity epoxy resin dry molding machine, conduct a power frequency-harmonic composite current temperature rise test, including: Step 6.2.1: Construct a high-voltage synthetic circuit by using the harmonic compensation method, and superimpose the 3rd, 5th and 7th characteristic harmonic currents onto the power frequency current.

[0375] This embodiment constructs a high-voltage synthetic circuit using the harmonic compensation method, superimposing the 3rd, 5th, and 7th characteristic harmonic currents onto the power frequency current, accurately simulating the actual operating conditions under renewable energy access scenarios.

[0376] Step 6.2.2: Use pre-embedded thermocouples and infrared thermal imagers to measure the temperature rise distribution of the windings, and use resistance strain gauges and Raman distributed fiber optic sensors to monitor changes in thermal stress.

[0377] This embodiment achieves multi-dimensional and accurate measurement of temperature rise and stress distribution by using pre-embedded thermocouples and infrared thermal imagers to measure the temperature rise distribution of the windings, and by using resistance strain gauges and Raman distributed fiber optic sensors to monitor changes in thermal stress.

[0378] Step 6.2.3: Run continuously until the temperature rise stabilizes, record temperature and stress data, establish the harmonic-temperature rise correspondence and thermal stress distribution model, and obtain the temperature rise and stress assessment results.

[0379] This embodiment continuously runs the machine until the temperature rise stabilizes, records temperature and stress data, establishes the harmonic-temperature rise correspondence and thermal stress distribution model, and obtains the temperature rise and stress evaluation results, providing complete data support for the thermal performance evaluation of the prototype.

[0380] Step 6.3: Perform a high-frequency oscillating pulse voltage inter-turn insulation test using the high thermal conductivity epoxy resin dry converter, including: Step 6.3.1: Use a high-voltage synthesis circuit with a trigger gap to generate an oscillating pulse voltage.

[0381] This embodiment uses a high-voltage synthetic circuit with a trigger gap to generate oscillating pulse voltage, which accurately simulates the overvoltage condition of power grid operation and creates real test conditions for inter-turn insulation performance testing.

[0382] Step 6.3.2: Measure the partial discharge initiation voltage by boosting the voltage at a rate of 1kV / s, and apply lightning impulse and switching impulse voltages.

[0383] This embodiment measures the partial discharge initiation voltage by boosting the voltage at a rate of 1kV / s, and applies lightning impulse and switching impulse voltages to systematically evaluate the insulation performance of the prototype under different overvoltage conditions.

[0384] Step 6.3.3: Monitor the discharge signal using a partial discharge detector, disassemble the coil after the test to observe the carbonization points, determine the insulation failure early warning threshold, and obtain the inter-turn insulation assessment results.

[0385] This embodiment monitors the discharge signal using a partial discharge detector, disassembles the coil after the test to observe the carbonization points, sets an insulation failure early warning threshold, obtains the inter-turn insulation evaluation results, and establishes a complete insulation performance evaluation system.

[0386] Step 6.3.4: Based on the temperature rise and stress assessment results and the inter-turn insulation assessment results, comprehensively determine whether the high thermal conductivity epoxy resin dry transformer prototype has passed the test verification: If the test verification is passed, the high thermal conductivity epoxy resin dry deformation sampler shall be regarded as the high thermal conductivity epoxy resin dry deformation sampler verified by the test. If the test verification fails, the specific failure mode of the high thermal conductivity epoxy resin dry-type transformer prototype will be located based on the failure to meet the temperature rise and stress assessment results and / or the failure to meet the inter-turn insulation assessment results, and the global design parameters of the dry-type transformer will be optimized accordingly.

[0387] This embodiment comprehensively determines whether the high thermal conductivity epoxy resin dry-type transformer prototype passes the test verification based on the temperature rise and stress assessment results and the inter-turn insulation assessment results. For those that fail, it identifies the specific failure mode and provides feedback to optimize the global design parameters of the dry-type transformer, thus forming a complete "design-manufacturing-testing-optimization" technical closed loop.

[0388] Compared with the prior art, the beneficial effects of this embodiment are as follows: 1. By using a multi-scale collaborative system of microscale molecular dynamics modeling, mesoscale phase field method modeling and macroscale finite element modeling, combined with orthogonal iterative method to optimize parameters, the defects of existing technology material formulation and structural design are overcome, and the precise linkage of microscale interface parameters, mesoscale structural characteristics and macroscale performance of high thermal conductivity epoxy resin is realized, which significantly improves the scientific nature and targeting of material design. 2. High-resolution defect data is obtained based on broadband terahertz detection technology. The mapping relationship between defect features and aging state is constructed by combining deep learning algorithms. This overcomes the shortcomings of traditional material testing, which makes it difficult to quantitatively identify internal microcracks and aging degree. It provides accurate basis for material design parameter correction and performance dataset establishment, and ensures the basic reliability of equipment design. 3. A multi-physics field verification method for high thermal conductivity epoxy resin dry-type transformers was constructed. Based on the complex operating conditions of renewable energy, a multi-physics field coupled simulation model was built. With the goals of maximizing capacity and minimizing volume, the Monte Carlo iteration method and global optimization algorithm were used to achieve multi-objective collaborative optimization. This method overcomes the limitations of single-physics field verification and local optimization in existing technologies. It achieves the optimal configuration of global design parameters such as dry-type transformer core structure, coil arrangement, and air duct layout, while taking into account both equipment compactness and high performance. 4. By simulating complex current conditions through power frequency-harmonic composite current temperature rise test, and combining it with high frequency oscillating pulse voltage inter-turn insulation test to cover overvoltage scenarios, the shortcomings of traditional test verification scenarios being singular and disconnected from actual working conditions are overcome. The temperature rise characteristics, thermal stress stability and insulation reliability of dry transformers are comprehensively verified, providing a solid guarantee for the large-scale application of prototypes.

[0389] Furthermore, in step S4, a high thermal conductivity epoxy resin composite material is used as the insulating and adhesive medium to wind a coil with radial heat dissipation channels. Specifically: During the coil winding process, strip-shaped supports are placed at predetermined positions between adjacent coil layers. The strip-shaped supports are made of volatile materials. End baffles are installed at both ends of the wound coil along the axial direction. The inner ring surface of the end baffles is provided with limiting grooves that are adapted to the position and shape of the strip-shaped support. The high thermal conductivity epoxy resin composite material is used to vacuum pressure impregnate the coil that has been wound and has end baffles installed, so that the high thermal conductivity epoxy resin composite material fills the gaps between coils and turns. The coil after vacuum pressure impregnation is subjected to a first-stage curing treatment. The temperature of the first-stage curing treatment is controlled within the range that allows the high thermal conductivity epoxy resin composite material to initially gel but is below the sublimation point of the volatile material. The coil, after the first stage of curing, is subjected to a second stage of gradient temperature curing. The temperature curve of the second stage of gradient temperature curing is set to allow the volatile material to completely sublimate and escape, forming a continuous radial heat dissipation channel between the coil layers, while simultaneously allowing the high thermal conductivity epoxy resin composite material to be completely cured and molded, forming a coil with radial heat dissipation channels.

[0390] Furthermore, in step S4, a high thermal conductivity epoxy resin composite material is used to bond and cure the stacked iron cores to form an integral iron core. Specifically: Multiple stamped iron chips are stacked according to a preset stacking diagram to form a stack body, so that the micro protrusions of adjacent iron chips are staggered and arranged to form a connected resin microchannel mesh inside the stack body. The iron chips have multiple micro protrusions protruding in the stacking direction at preset positions on their edges. In a vacuum environment, a high thermal conductivity epoxy resin composite material preheated to a preset viscosity range is injected and impregnated into the stack, so that the high thermal conductivity epoxy resin composite material fills the gaps between all iron chips through the resin microchannel mesh. Axial pressure is applied to the impregnated stack and gradient temperature is increased for curing. During the gradient temperature increase curing process, the thickness of the insulating layer between the iron chips is controlled and maintained uniformly through the mutual support of the micro bosses. After gradient heating and curing, the integral iron core is obtained through cooling and post-treatment.

[0391] Furthermore, the process of applying axial pressure to the impregnated stack and performing gradient temperature curing, wherein the mutual support of the micro-boobs controls and maintains the uniformity of the insulating layer thickness between the iron chips during the gradient temperature curing process, includes: In a vacuum environment, a preset first axial pressure is applied to the impregnated stack along its stacking direction and maintained for a first preset time, so that excess high thermal conductivity epoxy resin composite material is discharged from the resin microchannel mesh at the edge of the stack. After debinding, the stacked body is transferred to a curing mold, a second axial pressure higher than the first axial pressure is applied, and the stacked body is heated to a first preset temperature at a first heating rate, so that the high thermal conductivity epoxy resin composite material reaches a predetermined viscosity and begins to gel. The temperature is raised to a second preset temperature at a second heating rate and held for a second preset time. The change in the thickness of the stack is monitored in real time by the displacement sensor built into the curing mold, and the second axial pressure is dynamically adjusted based on the change to compensate for the shrinkage of the high thermal conductivity epoxy resin composite material during the curing process and maintain the uniform thickness of the insulation layer between the iron chips.

[0392] Further, in step S4, the coil is fitted onto the integral iron core to form an assembly, and an insulating component is installed on the assembly to form an axial heat dissipation channel. Temperature and stress sensors are pre-embedded inside the coil. Specifically: After covering the surface of the core column of the integral iron core with an insulating film, elastic insulating support strips with self-centering function are installed. The elastic insulating support strips are distributed along the core column axis of the integral iron core and arranged in a circumferential array. After the coil is heated to the preset expansion temperature, it is fitted onto the core post of the integral iron core on which the elastic insulating support bar has been installed. After the coil cools and shrinks, the inner wall of the coil is interference-fitted with the elastic insulating support bar and pressed tightly to form the first positioning. Insulating end rings are fitted at both ends of the coil along the axial direction, and U-shaped groove insulating partitions made of high thermal conductivity epoxy resin composite material are inserted between adjacent coils and between the coil and the overall iron core yoke. Each U-shaped groove insulating partition and the elastic insulating support strip are interleaved to form an axial heat dissipation channel extending along the coil axis. Before mounting the coil, the temperature and stress sensors are attached to the preset hot spots and stress concentration areas on the inner wall of the coil. The leads of the temperature and stress sensors are led out along the radial heat dissipation channel and temporarily fixed at the end of the coil through a detachable sealing joint.

[0393] Furthermore, in step S4, the assembly is integrally cast using a high thermal conductivity epoxy resin composite material in a vacuum environment, and then cured under gradient temperature control to produce a transformer prototype. Specifically: Before casting, the assembly is vacuum dried and the leads of the temperature and stress sensors are led out through the sealed terminals pre-set on the casting mold. The high thermal conductivity epoxy resin composite material, preheated to the casting viscosity, is injected into the casting mold in two stages under vacuum: In the first stage, the high thermal conductivity epoxy resin composite material is injected from the bottom of the casting mold at the first injection rate, so that the high thermal conductivity epoxy resin composite material preferentially fills the gaps between the coil, the integral iron core and the insulating parts at the bottom of the assembly. In the second stage, the high thermal conductivity epoxy resin composite material is injected from the top of the casting mold at a second injection rate higher than the first injection rate until the assembly is completely submerged in the high thermal conductivity epoxy resin composite material. After the casting is completed, the first negative pressure value is maintained in a vacuum environment and the mixture is left to stand for a first preset time to allow the air bubbles in the high thermal conductivity epoxy resin composite material to escape. Gradient temperature-controlled curing is performed on the casting mold and its internal materials after casting. The temperature of the casting mold and its internal material is raised to the first curing temperature at a first heating rate and kept at that temperature to allow the high thermal conductivity epoxy resin composite material to initially gel. The temperature of the casting mold and the internal material is raised to the second curing temperature at a second heating rate lower than the first heating rate and kept at the temperature to allow the high thermal conductivity epoxy resin composite material to fully cure. The temperature of the casting mold and the internal material is then slowly cooled to the demolding temperature at a controlled cooling rate. During the heat preservation stage of gradient temperature-controlled curing, the heat preservation temperature or heat preservation time is adjusted based on the real-time monitoring of temperature and stress changes of the internal material using temperature and stress sensors led out from the sealed terminal.

[0394] Furthermore, the transformer prototype testing method described in step S4 includes: An oscillating pulse voltage wave simulating frequent overvoltage conditions is applied to the high-voltage winding of the transformer prototype, while a power frequency voltage is applied to the low-voltage winding. The inter-turn insulation withstand data of the transformer prototype are continuously acquired and recorded at a preset period. A composite current containing the fundamental wave and a specified harmonic is applied to a transformer prototype to simulate load fluctuations caused by renewable energy access. Temperature rise distribution, hot spot data and mechanical vibration spectrum of the transformer prototype are collected and generated by pre-embedded temperature and stress sensors. Based on the inter-turn insulation tolerance data, temperature rise distribution, hot spot data, and mechanical vibration spectrum, the thermal conductivity, insulation performance, and mechanical performance of the transformer prototype are evaluated. The transformer prototype was manufactured using the aforementioned method.

[0395] Furthermore, the step of applying oscillating pulse voltage waves simulating frequent overvoltage conditions to the high-voltage winding of the transformer prototype, while simultaneously applying power frequency voltage to the low-voltage winding, and continuously acquiring and recording the inter-turn insulation withstand data of the transformer prototype for a preset period, includes: The transformer prototype was placed in a shielded test environment, and the first end of the high voltage winding was connected to the output end of the high voltage pulse generator. The first and last ends of the low voltage winding were short-circuited and connected to one pole of the power frequency voltage source. At the same time, the core and shell of the transformer prototype were grounded. An oscillating pulse voltage wave is applied to the high-voltage winding at a preset repetition frequency and pulse amplitude. The waveform of the oscillating pulse voltage wave is a decaying oscillating waveform, the voltage value of its first peak is a preset multiple of the peak value of the rated phase voltage of the transformer prototype, and its oscillation frequency is within a preset range of the inherent resonant frequency of the transformer prototype winding. While applying an oscillating pulse voltage wave to the high-voltage winding, a power frequency sinusoidal voltage equal to the rated voltage of the transformer prototype is applied to the low-voltage winding, and the phase of the power frequency sinusoidal voltage is controlled so that its peak voltage moment is synchronized with the first peak moment of the oscillating pulse voltage wave. A complete test sequence is formed by continuously applying voltage combinations a preset number of times. In each test sequence, the pulse impact current waveform and partial discharge signal of the high-voltage winding, as well as the power frequency leakage current of the low-voltage winding, are collected synchronously. The transfer function of the winding is calculated based on the pulse impact current waveform, and the inter-turn insulation withstand data of the transformer prototype is analyzed and recorded according to the changes in the partial discharge signal and the power frequency leakage current.

[0396] Furthermore, the application of a composite current containing the fundamental wave and a specified harmonic to the transformer prototype to simulate load fluctuations from renewable energy integration, and the acquisition and generation of temperature rise distribution, hotspot data, and mechanical vibration spectrum of the transformer prototype through pre-embedded temperature and stress sensors, includes: A test current, composed of a power frequency fundamental wave and a specified harmonic with a preset amplitude and phase, is applied to the low-voltage winding of the transformer prototype using a programmable composite current source to simulate a typical harmonic spectrum with a power electronic converter connected. The programmable composite current source is controlled to output the test current according to a preset load cycle curve. In the load cycle curve, the amplitude of the fundamental current changes periodically, and the content of each harmonic is adjusted according to a preset ratio to simulate the load fluctuation of renewable energy access. During the application of the test current, data from temperature and stress sensors embedded inside the coil are collected simultaneously, and vibration acceleration signals are collected at preset monitoring points on the outer shell of the transformer prototype. The collected temperature data is mapped and fitted to the three-dimensional structural model of the transformer prototype to generate a dynamic temperature rise distribution map during the operation of the transformer prototype, and the temperature area exceeding the preset threshold is identified as hot spot data. Time-frequency analysis was performed on the collected stress data and vibration acceleration signals to extract characteristic frequency components related to the specified harmonic frequency, and the mechanical vibration spectrum of the transformer prototype under harmonic excitation was synthesized.

[0397] Furthermore, the evaluation of the thermal conductivity, insulation performance, and mechanical performance of the transformer prototype based on inter-turn insulation tolerance data, temperature rise distribution, hot spot data, and mechanical vibration spectrum includes: Based on the dynamic temperature rise distribution map and the hot spot data, the thermal conductivity of the transformer prototype is evaluated using computer-equivalent thermal network parameters. Based on the inter-turn insulation tolerance data, a quantitative index of insulation aging degree is calculated to evaluate the insulation performance of the transformer prototype. The vibration spectrum of the structure is compared with the theoretical spectrum obtained by modal simulation based on the three-dimensional model of the transformer prototype to evaluate the mechanical performance of the transformer prototype.

[0398] For example: This embodiment describes the implementation steps of a transformer prototype manufacturing method, including: Step 1: Using a high thermal conductivity epoxy resin composite material as the insulation and bonding medium, a coil with radial heat dissipation channels is wound.

[0399] In this embodiment, a high thermal conductivity epoxy resin composite material is used as the insulation and bonding medium to wind a coil with radial heat dissipation channels. Through the high thermal conductivity of the material and the structural design of the heat dissipation channels, the heat dissipation efficiency of the coil is significantly improved, the temperature rise gradient between coils is reduced, and the insulation performance and structural stability are guaranteed.

[0400] Step 1.1: During the coil winding process, strip-shaped supports are placed at predetermined positions between adjacent coil layers. The strip-shaped supports are made of volatile materials.

[0401] In this embodiment, during the coil winding process, strip-shaped supports made of volatile materials are placed at preset positions between adjacent coil layers. The temporary support of the supports ensures that the winding gap is uniform, providing a spatial basis for the subsequent sublimation and escape of volatile materials to form a continuous radial heat dissipation channel, and avoiding coil displacement or deformation during the winding process.

[0402] Step 1.2: Install end baffles at both ends of the wound coil along the axial direction. The inner ring surface of the end baffles is provided with limiting grooves that are adapted to the position and shape of the strip support.

[0403] In this embodiment, end baffles are installed at both ends of the wound coil along the axial direction. The inner ring surface of the end baffles is provided with limiting grooves that are adapted to the position and shape of the strip-shaped support. The position of the support is fixed by physical limiting, ensuring that the support does not shift during the vacuum pressure impregnation process, and ensuring the positional accuracy and continuity of the radial heat dissipation channel.

[0404] Step 1.3: Vacuum pressure impregnation is performed on the coil that has been wound and has end baffles installed using the high thermal conductivity epoxy resin composite material, so that the high thermal conductivity epoxy resin composite material fills the gaps between coils and turns.

[0405] In this embodiment, a high thermal conductivity epoxy resin composite material is used to vacuum pressure impregnate the coil that has been wound and has end baffles installed. The vacuum environment eliminates air between the coils and turns, and the pressure makes the composite material fully fill the gaps, thereby improving the insulation strength and thermal conductivity uniformity of the coil and reducing the risk of partial discharge.

[0406] Step 1.4: Perform a first-stage curing treatment on the vacuum pressure impregnated coil. The temperature of the first-stage curing treatment is controlled within the range that allows the high thermal conductivity epoxy resin composite material to initially gel but is below the sublimation point of the volatile material.

[0407] In this embodiment, the coil after vacuum pressure impregnation is subjected to a first-stage curing treatment. The temperature is controlled within the range that allows the high thermal conductivity epoxy resin composite material to initially gel but is below the sublimation point of the volatile material. This ensures that the composite material is initially shaped and does not flow, while also preventing the premature volatilization of the volatile material from damaging the air passage structure, thus creating conditions for the second-stage curing.

[0408] Step 1.5: The coil after the first stage of curing is subjected to a second stage of gradient temperature curing. The temperature curve of the second stage of gradient temperature curing is set to allow the volatile material to completely sublimate and escape, forming a continuous radial heat dissipation channel between the coil layers, while allowing the high thermal conductivity epoxy resin composite material to be completely cured and formed into a coil with radial heat dissipation channels.

[0409] In this embodiment, the coil that has been cured in the first stage is subjected to a second stage of gradient temperature curing. By setting the temperature curve, the volatile materials are completely sublimated and escaped, forming a continuous radial heat dissipation channel between the coil layers. At the same time, the high thermal conductivity epoxy resin composite material is completely cured and molded, achieving the dual effect of improved heat dissipation efficiency and enhanced structural strength, thus extending the service life of the coil.

[0410] Step 2: Use high thermal conductivity epoxy resin composite material to bond and cure the stacked iron cores to form an integral iron core.

[0411] In this embodiment, a high thermal conductivity epoxy resin composite material is used to bond and cure stacked iron cores to form an integral iron core. Through the thermal conductivity characteristics of the high thermal conductivity material and the resin microchannel mesh formed by the staggered arrangement of micro bosses, the internal heat conduction path of the iron core is optimized and the insulation layer thickness is uniformly controlled, thereby improving the thermal uniformity, mechanical stability and resistance to local overheating of the iron core.

[0412] Step 2.1: Stack multiple stamped iron chips according to a preset stacking diagram to form a stack body, so that the micro protrusions of adjacent iron chips are staggered and arranged to form a connected resin microchannel mesh inside the stack body. The iron chips have multiple micro protrusions protruding in the stacking direction at preset positions on their edges.

[0413] In this embodiment, multiple stamped iron chips are stacked according to a preset stacking diagram to form a stack body. The micro protrusions of adjacent iron chips are staggered to form a connected resin microchannel mesh inside the stack body. The staggered protrusion design provides uniform penetration channels for the composite material, ensuring that the resin can completely fill the gaps between the iron chips during impregnation. At the same time, a thermally conductive microchannel network is formed, which enhances the thermal conductivity of the overall iron core.

[0414] Step 2.2: In a vacuum environment, inject and impregnate the stack with a high thermal conductivity epoxy resin composite material preheated to a preset viscosity range, so that the high thermal conductivity epoxy resin composite material fills the gaps between all iron chips through the resin microchannel mesh.

[0415] In this embodiment, a high thermal conductivity epoxy resin composite material is injected and impregnated under vacuum and preheated to a preset viscosity. By eliminating air through the vacuum environment and utilizing the resin microchannel mesh, the composite material fully fills all the gaps between the iron cores, improving the insulation strength and thermal conductivity uniformity of the iron core and reducing the risk of partial discharge and thermal accumulation.

[0416] Step 2.3: Apply axial pressure to the impregnated stack and perform gradient temperature curing. During the gradient temperature curing process, the thickness of the insulating layer between the iron chips is controlled and maintained uniformly through the mutual support of the micro bosses.

[0417] In this embodiment, axial pressure is applied to the impregnated stack and gradient temperature is increased for curing. The mutual support of the micro bosses controls and maintains the uniform thickness of the insulation layer between the iron chips, avoiding uneven thickness caused by material shrinkage during the curing process, and ensuring the consistency of the mechanical strength and thermal conductivity of the iron core.

[0418] Step 2.3.1: In a vacuum environment, apply a preset first axial pressure to the impregnated stack along its stacking direction and maintain it for a first preset time to allow excess high thermal conductivity epoxy resin composite material to be discharged from the resin microchannel mesh at the edge of the stack.

[0419] In this embodiment, a preset first axial pressure is applied along the stacking direction in a vacuum environment and maintained for a first preset time, so that excess composite material is discharged from the resin microchannel mesh, reducing internal voids and bubbles, and improving the density and thermal conductivity of the iron core.

[0420] Step 2.3.2: Transfer the desorbed stack to the curing mold, apply a second axial pressure higher than the first axial pressure, and heat the stack to the first preset temperature at the first heating rate, so that the high thermal conductivity epoxy resin composite material reaches the predetermined viscosity and begins to gel.

[0421] In this embodiment, the stacked body after glue removal is transferred to a curing mold to apply a higher second axial pressure, and heated to a first preset temperature at a first heating rate to gel the composite material. Through the coordinated control of pressure and temperature, the initial shaping and structural stability of the material are ensured.

[0422] Step 2.3.3: Raise the temperature to the second preset temperature at the second heating rate and hold it at the second preset time. Monitor the change in the thickness of the stack in real time through the displacement sensor built into the curing mold, and dynamically adjust the second axial pressure based on the change to compensate for the shrinkage of the high thermal conductivity epoxy resin composite material during the curing process and maintain the uniform thickness of the insulation layer between the iron chips.

[0423] This embodiment uses a displacement sensor to monitor the thickness change of the stack in real time and dynamically adjust the second axial pressure to compensate for the curing shrinkage of the composite material, maintain the uniform thickness of the insulating layer between the iron chips, and avoid stress concentration and performance degradation caused by shrinkage.

[0424] Step 2.4: After gradient heating and curing, the integral iron core is obtained through cooling and post-treatment.

[0425] This embodiment is based on obtaining an integral iron core after gradient heating and curing, followed by cooling and post-processing. Through scientific curing process and post-processing, the final structural stability and performance reliability of the iron core are ensured.

[0426] Step 3: Fit the coil onto the integral iron core to form an assembly, and install insulating parts on the assembly to form an axial heat dissipation channel. Pre-embed temperature and stress sensors inside the coil.

[0427] In this embodiment, the coil is fitted onto the integral iron core to form an assembly, and insulating components are installed to form an axial heat dissipation channel. With the help of pre-embedded temperature and stress sensors, the assembly achieves efficient heat dissipation, precise positioning, and real-time status monitoring, thereby improving the reliability of transformer operation and supporting fault early warning.

[0428] Step 3.1: After covering the surface of the core column of the integral iron core with an insulating film, install elastic insulating support strips with self-centering function. The elastic insulating support strips are distributed along the axial direction of the core column of the integral iron core and arranged in a circumferential array.

[0429] In this embodiment, after covering the surface of the entire iron core column with an insulating film, an elastic insulating support bar with a self-centering function is installed. The self-centering function of the support bar ensures the accurate positioning of the coil assembly, and at the same time provides structural support for the subsequent formation of axial heat dissipation channels.

[0430] Step 3.2: After heating the coil to the preset expansion temperature, it is fitted onto the core post of the integral iron core on which the elastic insulating support bar has been installed. After the coil cools and shrinks, the inner wall of the coil and the elastic insulating support bar are interference-fitted and tightly pressed together to form the first positioning.

[0431] In this embodiment, the coil is heated to a preset expansion temperature and then fitted onto the iron core column that has been fitted with elastic insulating support bars. After cooling, a tight press-fit is formed through interference fit, which realizes a reliable connection and precise positioning between the coil and the iron core, and enhances the structural stability.

[0432] Step 3.3: Install insulating end rings on both ends of the coil along the axial direction, and insert U-shaped groove insulating partitions made of high thermal conductivity epoxy resin composite material between adjacent coils and between the coil and the overall iron core yoke. Each U-shaped groove insulating partition and the elastic insulating support strip are interleaved to form an axial heat dissipation channel extending along the coil axis.

[0433] In this embodiment, insulating end rings are fitted at both ends of the coil along the axial direction and U-shaped groove insulating partitions are inserted. The partitions and elastic insulating support strips are interlocked to form an axial heat dissipation channel, which significantly improves the axial heat dissipation efficiency of the assembly and reduces the hot spot temperature.

[0434] Step 3.4: Before mounting the coil, attach the temperature and stress sensor to the preset hot spots and stress concentration areas on the inner wall of the coil. The leads of the temperature and stress sensor are led out along the radial heat dissipation channel and temporarily fixed at the end of the coil through a detachable sealing joint.

[0435] In this embodiment, temperature and stress sensors are attached to hot spots and stress concentration areas on the inner wall of the coil before the coil is assembled. The leads are led out along the radial heat dissipation channel and fixed by a detachable sealing joint, which realizes real-time monitoring of temperature and stress in key areas and supports operation status assessment and fault early warning.

[0436] Step 4: The assembly is cast as a whole in a vacuum environment using a high thermal conductivity epoxy resin composite material, and then cured under gradient temperature control to form a transformer prototype.

[0437] In this embodiment, a high thermal conductivity epoxy resin composite material is used to cast the assembly as a whole in a vacuum environment and then cured with gradient temperature control. By eliminating air bubbles in a vacuum environment and injecting in stages to ensure uniform material filling, combined with the gradient temperature control curing process, high-quality molding and performance optimization of the transformer prototype are achieved.

[0438] Step 4.1: Before casting, the assembly is vacuum dried and the leads of the temperature and stress sensors are led out through the sealed terminals pre-set on the casting mold.

[0439] In this embodiment, the assembly is vacuum dried before casting and the sensor leads are led out through sealed terminals to eliminate internal moisture and air bubbles, ensuring the safety and electrical reliability of the casting process.

[0440] Step 4.2: The high thermal conductivity epoxy resin composite material, preheated to the casting viscosity, is injected into the casting mold in two stages under vacuum: In the first stage, the high thermal conductivity epoxy resin composite material is injected from the bottom of the casting mold at the first injection rate, so that the high thermal conductivity epoxy resin composite material preferentially fills the gaps between the coil, the integral iron core and the insulating parts at the bottom of the assembly. In the second stage, the high thermal conductivity epoxy resin composite material is injected from the top of the casting mold at a second injection rate higher than the first injection rate until the assembly is completely submerged.

[0441] This embodiment involves injecting the composite material in two stages. In the first stage, the material is injected from the bottom to fill the bottom gaps first. In the second stage, the material is injected from the top to completely immerse the assembly. This staged injection ensures uniform material distribution and full filling of gaps, thereby improving insulation and thermal conductivity.

[0442] Step 4.3: After casting is completed, maintain the first negative pressure value in a vacuum environment and let it stand for a first preset time to allow the air bubbles in the high thermal conductivity epoxy resin composite material to escape.

[0443] After casting is completed in this embodiment, the sample is kept under negative pressure in a vacuum environment to allow air bubbles in the composite material to escape, reducing internal defects and improving the mechanical strength and insulation reliability of the prototype.

[0444] Step 4.4: Perform gradient temperature-controlled curing on the casting mold and its internal materials after pouring. The temperature of the casting mold and its internal material is raised to the first curing temperature at a first heating rate and kept at that temperature to allow the high thermal conductivity epoxy resin composite material to initially gel. The temperature of the casting mold and the internal material is raised to the second curing temperature at a second heating rate lower than the first heating rate and held at that temperature to allow the high thermal conductivity epoxy resin composite material to fully cure. The temperature of the casting mold and the internal material is then slowly cooled to the demolding temperature at a controlled cooling rate.

[0445] This embodiment employs a gradient temperature-controlled curing process. First, the material is heated to a first curing temperature at a first heating rate to allow it to initially gel. Then, it is heated to a second curing temperature at a second heating rate to fully cure it. Finally, the cooling rate is controlled to the demolding temperature, thus avoiding thermal stress damage and ensuring curing quality.

[0446] Step 4.5: During the heat preservation stage of gradient temperature-controlled curing, the heat preservation temperature or heat preservation time is adjusted according to the temperature and stress changes of the internal material monitored in real time, using temperature and stress sensors led out from the sealed terminal.

[0447] In this embodiment, during the gradient temperature-controlled curing and insulation stage, sensors are used to monitor temperature and stress changes in real time and adjust insulation parameters, thereby achieving closed-loop control of the curing process, optimizing the insulation curing quality and mechanical properties, and improving the overall performance and reliability of the transformer prototype.

[0448] The implementation steps of a transformer prototype testing method include: Step 1: Apply an oscillating pulse voltage wave simulating frequent overvoltage conditions to the high-voltage winding of the transformer prototype, and simultaneously apply a power frequency voltage to the low-voltage winding. Continuously acquire and record the inter-turn insulation withstand data of the transformer prototype for a preset period.

[0449] In this embodiment, an oscillating pulse voltage wave simulating frequent overvoltage conditions is applied to the high-voltage winding of the transformer prototype, while a power frequency voltage is applied to the low-voltage winding. The inter-turn insulation withstand data is acquired and recorded continuously for a preset period. The insulation stress under actual working conditions is accurately simulated through the dual-voltage combination test, providing direct data support for evaluating the insulation reliability of the winding under the combined action of overvoltage and power frequency voltage.

[0450] Step 1.1: Place the transformer prototype in a shielded test environment, connect the first end of the high-voltage winding to the output end of the high-voltage pulse generator, short-circuit the first and last ends of the low-voltage winding and connect them to one pole of the power frequency voltage source, and ground the core and shell of the transformer prototype.

[0451] In this embodiment, the transformer prototype is placed in a shielded test environment and the electrical connections of the high-voltage winding and low-voltage winding and the grounding of the core shell are completed. The shielded environment eliminates external interference, ensures the accuracy of voltage application and signal acquisition, and guarantees that the test conditions meet the standard requirements.

[0452] Step 1.2: Apply an oscillating pulse voltage wave to the high-voltage winding at a preset repetition frequency and pulse amplitude. The waveform of the oscillating pulse voltage wave is a decaying oscillating waveform, the voltage value of its first peak is a preset multiple of the peak value of the rated phase voltage of the transformer prototype, and its oscillation frequency is within a preset range of the inherent resonant frequency of the transformer prototype winding.

[0453] In this embodiment, a damped oscillating waveform voltage wave is applied with a preset repetition frequency and pulse amplitude. The voltage value of the first peak is a preset multiple of the peak value of the rated phase voltage and the oscillation frequency is within the range of the winding's inherent resonant frequency. By precisely controlling the pulse parameters, the insulation withstand capability of the winding can be specifically assessed, thereby improving the matching degree between the test results and the actual overvoltage conditions.

[0454] Step 1.3: While applying an oscillating pulse voltage wave to the high-voltage winding, apply a power frequency sinusoidal voltage equal to the rated voltage of the transformer prototype to the low-voltage winding, and control the phase of the power frequency sinusoidal voltage so that its peak voltage time is synchronized with the first peak time of the oscillating pulse voltage wave.

[0455] In this embodiment, while applying an oscillating pulse wave to the high-voltage winding, a power frequency sinusoidal voltage equal to the rated voltage is applied to the low-voltage winding, and its phase is controlled to be synchronized with the first peak of the pulse wave. By applying dual voltages synchronously, a composite electric stress scenario is simulated, accurately reproducing the electric field coupling effect between windings in actual operation, and improving the authenticity of insulation withstand data.

[0456] Step 1.4: Apply voltage combinations a preset number of times to form a complete test sequence, and in each test sequence, synchronously collect the pulse impact current waveform and partial discharge signal of the high voltage winding, as well as the power frequency leakage current of the low voltage winding.

[0457] In this embodiment, a complete test sequence is formed by continuously applying a preset number of voltage combinations. The pulse impact current of the high-voltage winding, the partial discharge signal, and the power frequency leakage current of the low-voltage winding are collected simultaneously. Through the synchronous acquisition of multiple parameters, the insulation status is comprehensively monitored, providing a multi-dimensional data source for the inter-turn insulation withstand capability analysis.

[0458] Step 1.5: Calculate the transfer function of the winding based on the pulse impact current waveform, and analyze and record the inter-turn insulation withstand data of the transformer prototype according to the changes in the partial discharge signal and the power frequency leakage current.

[0459] This embodiment calculates the winding transfer function based on the pulse impact current waveform, and combines partial discharge signal and power frequency leakage current change analysis to record the inter-turn insulation tolerance data. Through dual analysis of quantitative indicators and signal characteristics, it achieves accurate assessment of insulation aging degree and fault early warning.

[0460] Step 2: Apply a composite current containing the fundamental wave and a specified harmonic to the transformer prototype to simulate the load fluctuations of renewable energy access. Collect and generate the temperature rise distribution, hot spot data and mechanical vibration spectrum of the transformer prototype through pre-embedded temperature and stress sensors.

[0461] In this embodiment, a composite current containing the fundamental wave and a specified harmonic is applied to a transformer prototype to simulate the fluctuation of renewable energy load. Temperature rise distribution, hot spot data and mechanical vibration spectrum are collected by pre-embedded sensors to provide dynamic monitoring data for evaluating the thermo-mechanical performance of the transformer under harmonic excitation.

[0462] Step 2.1: Apply a test current, which is a superposition of the power frequency fundamental wave and a specified harmonic with a preset amplitude and phase, to the low-voltage winding of the transformer prototype through a programmable composite current source, so as to simulate the typical harmonic spectrum of a transformer with a power electronic converter.

[0463] This embodiment applies a test current superimposed with the fundamental frequency wave and a specified harmonic by a programmable composite current source, simulating the typical harmonic spectrum of a power electronic converter, accurately reproducing the current waveform characteristics under renewable energy access scenarios, and improving the representativeness of the test conditions.

[0464] Step 2.2: Control the programmable composite current source to output the test current according to the preset load cycle curve. In the load cycle curve, the amplitude of the fundamental current changes periodically, and the content of each harmonic is adjusted according to the preset ratio to simulate the load fluctuation of renewable energy access.

[0465] In this embodiment, the control current source outputs the test current according to the preset load cycle curve, the fundamental amplitude changes periodically and the harmonic content is dynamically adjusted. Through dynam...

Claims

1. A comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer, characterized in that, Includes the following steps: S1. A performance control method based on high thermal conductivity epoxy resin composite material is adopted. The performance of high thermal conductivity epoxy resin composite material is controlled according to the preset material performance target value to obtain optimized material design parameters. The material design parameters include epoxy resin matrix parameters, compound thermally conductive filler parameters, and interface modifier parameters. S2. Using a detection and evaluation method based on high thermal conductivity epoxy resin composite materials, the high thermal conductivity epoxy resin composite material samples prepared according to the material design parameters are detected and evaluated to obtain the sample evaluation results. Based on the sample evaluation results, the material design parameters are corrected to obtain the corrected material design parameters. S3. Using the high thermal conductivity epoxy resin dry transformer design and verification method, based on the modified material design parameters, the dry transformer undergoes multi-scale modeling, structural design, multi-physics field coupling simulation verification, and multi-objective collaborative optimization to obtain the optimized global design parameters of the dry transformer. S4. Using the transformer prototype manufacturing method, a high thermal conductivity epoxy resin-based dry-type transformer prototype is manufactured according to the optimized global design parameters of the dry-type transformer. The transformer prototype is then tested and verified using the transformer prototype testing method to complete the comprehensive preparation and performance assurance of the high thermal conductivity epoxy resin-based dry-type transformer. The performance control method is used to determine the filler combination, interface modifier, and related interface and structural parameters of the high thermal conductivity epoxy resin composite material, so as to achieve precise control of the macroscopic thermal conductivity, electrical strength, and flexural strength of the material; the detection and evaluation method is used to quantitatively detect the defect type, location, and size of the composite material sample, and evaluate the aging degree and remaining life of the sample; the dry-type transformer design and verification method is used to complete the microscopic, mesoscopic, and macroscopic multi-scale modeling of the dry-type transformer, as well as the structural design and multi-physics field coupling simulation verification of the dry-type transformer based on the renewable energy access scenario; the transformer prototype manufacturing method is used to complete the molding and overall casting and curing of the transformer coil and core; and the transformer prototype testing method is used to verify the inter-turn insulation, thermal conductivity, and mechanical properties of the transformer prototype.

2. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S1, based on the preset performance target value, the model parameter set is obtained by screening the filler combination scheme and the corresponding interface modifier through high-throughput calculation. Specifically: Based on a pre-set filler database, a performance prediction proxy model is trained using a machine learning algorithm. The inputs of the performance prediction proxy model include filler type, filler shape, filler volume fraction, and molecular descriptor of the interface modifier. Using the performance target value as the optimization objective, a multi-objective optimization algorithm is used to perform a global search in the prediction space of the performance prediction proxy model to select the Pareto optimal solution set. By extracting common features from the Pareto optimal solution set, the filler combination scheme and the corresponding interface modifier are determined, forming a model parameter set.

3. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 2, characterized in that, The process of determining the filler combination scheme and corresponding interface modifier by extracting common features from the Pareto optimal solution set, forming a model parameter set, includes: The Pareto optimal solution set is grouped using a clustering algorithm. The characteristic variables of the clustering algorithm include filler type, filler shape, filler gradation ratio, and molecular descriptor of interface modifier. Based on each characteristic variable, a characteristic statistical matrix is ​​established to quantitatively calculate the average aspect ratio of the filler, the variance of the aspect ratio distribution, the gradation ratio of the filler, and the numerical concentration interval of the molecular descriptor of the interface modifier. Based on the feature statistics matrix, principal component analysis is used to identify the feature combinations that affect the performance of high thermal conductivity epoxy resin composites, and filler combination schemes that appear repeatedly in multiple Pareto optimal solutions and whose frequency exceeds a preset threshold are selected. The structural parameters of the interface modifier most strongly correlated with the filler combination scheme were determined by correlation analysis. The structural parameters include molecular chain length, type and number of polar groups, and molecular conformation characteristics. A model parameter set is constructed based on the selected filler combination schemes and the structural parameters of their corresponding interface modifiers.

4. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S1, the optimal interface parameters are obtained by calculating the interface binding energy and interface thermal conductivity based on the model parameter set through molecular dynamics simulation. Specifically: Based on the structural parameters of each filler combination scheme and its corresponding interface modifier in the model parameter set, a three-component atomic model including epoxy resin matrix, filler surface and interface modifier is constructed. Energy minimization and molecular dynamics relaxation simulations were performed sequentially on each three-component atomic model to obtain a stable equilibrium system structure. Based on the aforementioned equilibrium system structure, the interfacial binding energy corresponding to each filler combination scheme is calculated using energy analysis methods in molecular dynamics simulations. The interfacial binding energy is used to characterize the adsorption strength between the interfacial modifier and the filler surface. Based on the non-equilibrium molecular dynamics method, a temperature gradient is established at both ends of the equilibrium system structure, and the interfacial thermal conductivity corresponding to each packing combination scheme is calculated by the steady-state heat flow method. Based on the calculation results of interfacial binding energy and interfacial thermal conductivity corresponding to all filler combination schemes, quantitative correlation models between the molecular structure parameters of the interfacial modifier and the interfacial binding energy and interfacial thermal conductivity were established by multiple regression analysis. Based on the quantitative correlation model, a multi-objective optimization algorithm is used to iteratively optimize the molecular structure parameters of the interface modifier with the goal of simultaneously maximizing the interfacial binding energy and interfacial thermal conductivity, so as to obtain the optimal interfacial parameters.

5. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 4, characterized in that, Based on the calculation results of interfacial binding energy and interfacial thermal conductivity corresponding to all filler combination schemes, quantitative correlation models between the molecular structure parameters of the interfacial modifier and the interfacial binding energy and interfacial thermal conductivity are established through multiple regression analysis, including: The molecular structure parameters of the interface modifier are used as the set of independent variables, and the interface binding energy and interface thermal conductivity are used as the dependent variables, respectively, to construct an initial multivariate regression dataset. The initial multivariate regression dataset is standardized, and the variance inflation factor method is used to diagnose multicollinearity. Independent variables with variance inflation factors greater than a preset threshold are removed to obtain the processed dataset. Based on the processed dataset, the stepwise regression method was used to screen the molecular structure parameters of the interface modifiers with the p-value of the statistical test being less than a preset threshold, and an initial quantitative correlation model was established. The prediction accuracy of the initial quantitative correlation model was evaluated using k-fold cross-validation. Based on the evaluation results, the coefficients of the initial quantitative correlation model were regularized and optimized using ridge regression algorithm to obtain a quantitative correlation model between the molecular structure parameters of the interface modifier and the interface binding energy and interface thermal conductivity.

6. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 4, characterized in that, In step S1, based on the optimal interface parameters, the mesoscopic structural characteristic parameters are obtained by simulating the packing dispersion process and interface layer evolution process using the phase field method. Specifically: Based on the optimal interface parameters, a three-phase phase field model is established, which includes an epoxy resin matrix phase, a filler phase, and an interface modifier phase. Three sets of sequence parameters are defined to characterize the distribution state of the epoxy resin matrix phase, the filler phase, and the interface modifier phase, respectively. Based on the interface bonding energy and interface thermal conductivity in the optimal interface parameters, the interface energy parameters and gradient energy coefficients between each phase in the three-phase phase field model are set. Run the three-phase phase field model until the system reaches a dynamic equilibrium state; The spatial distribution, orientation distribution, packing network shape, and interface layer thickness distribution of the packing under dynamic equilibrium conditions are extracted as simulation results. Based on the simulation results, the permeation threshold, thermal conductivity path density, average thickness of the interface layer and its distribution uniformity of the packing network are quantitatively calculated to form a set of mesoscopic structural feature parameters that include packing distribution characteristics and interface layer characteristics.

7. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 6, characterized in that, Running the three-phase phase field model until the system reaches a dynamic equilibrium state includes: In the three-phase phase field model, a convection term describing the shear flow field is introduced. By coupling the flow field control equation and the phase field evolution equation, the dispersion process, migration process and agglomeration kinetics of the filler in the epoxy resin matrix are simulated until the system reaches a dynamic equilibrium state. Based on the optimal interface parameters, the interface gradient parameters are set, and the adsorption process of the interface modifier on the filler surface and the formation and evolution process of the interface layer are simulated by solving the interface evolution equation until the system reaches a dynamic equilibrium state.

8. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 7, characterized in that, The convection term describing the shear flow field is introduced into the three-phase phase field model. By coupling the flow field control equations and the phase field evolution equations, the dispersion, migration, and agglomeration dynamics of the filler in the epoxy resin matrix are simulated until the system reaches a dynamic equilibrium state, including: Based on the actual processing parameters, the flow field parameters in the three-phase phase field model are initialized, including the shear rate, viscosity and density of the epoxy resin matrix; Establish the coupled flow field control equations and phase field evolution equations; A convection term describing the shear flow field is introduced into the phase field evolution equation. The convection term is composed of the product of the flow field velocity vector and the order parameter gradient, and is used to describe the influence of the flow field on the packing transport and interface evolution. The coupled equations are spatially discretized using the finite difference method, and time integration is performed using an implicit-explicit hybrid time-progression method. Within each time step, the flow field control equations are first solved to obtain the updated flow field velocity distribution. Then, the flow field velocity distribution is substituted as a known quantity into the phase field evolution equations to solve for the order parameters of each time step. Based on the sequence parameters at each time step, the dispersion, migration and aggregation kinetics of fillers in the epoxy resin matrix are simulated by an iterative method until the system reaches a dynamic equilibrium state. The process of setting interface gradient parameters based on the optimal interface parameters, simulating the adsorption process of the interface modifier on the filler surface and the formation and evolution of the interface layer by solving the interface evolution equation, until the system reaches a dynamic equilibrium state, includes: Based on the interfacial binding energy in the optimal interfacial parameters, the adsorption energy parameters of the interfacial modifier on the filler surface are determined, and the interfacial gradient energy coefficient in the three-phase phase field model is set based on the adsorption energy parameters. Initialize the concentration distribution field of the interface modifier in the epoxy resin matrix and set adsorption boundary conditions on the filler surface; A non-conservative Allen-Cahn equation describing the evolution of the interface layer is established. The non-conservative Allen-Cahn equation includes an interface gradient term, a double-well potential function term, and a convection transport term. Based on the interfacial gradient energy coefficient and the adsorption energy parameter, the interfacial evolution equation is solved to simulate the adsorption process of the interfacial modifier on the filler surface and the formation and evolution process of the interfacial layer until the system reaches a dynamic equilibrium state.

9. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S1, the macroscopic thermal conductivity, electrical strength, and flexural strength of the high thermal conductivity epoxy resin composite material are predicted using finite element analysis based on the mesoscopic structural characteristic parameters, resulting in predicted performance values. Specifically: Based on the filler spatial distribution, orientation distribution, and filler network shape in the mesoscopic structural characteristic parameters, a three-dimensional geometric model of the high thermal conductivity epoxy resin composite material is reconstructed. Based on the filler type and the thickness distribution of the interface layer, the filler phase, the interface layer and the epoxy resin matrix phase are respectively assigned corresponding material properties, including thermal conductivity, dielectric constant, electrical conductivity and elastic modulus; The three-dimensional geometric model is meshed, and finite element models of the heat conduction control equation, the electrostatic conduction control equation, and the linear elasticity control equation are established respectively. Temperature boundary conditions are applied to the finite element model of heat conduction, and the temperature field distribution is obtained by solving. The macroscopic thermal conductivity of the high thermal conductivity epoxy resin composite material is calculated based on Fourier's law. Voltage boundary conditions are applied to the finite element model of electrostatic conduction, and the electric field distribution is obtained by solving. The electrical strength of the high thermal conductivity epoxy resin composite material is predicted based on the extreme values ​​of the electric field strength. Displacement and stress boundary conditions are applied in the finite element model of online elasticity, and the stress-strain field is obtained by solving. The bending strength of high thermal conductivity epoxy resin composite material is predicted based on the maximum stress criterion. The macroscopic thermal conductivity, electrical strength, and flexural strength are used as performance prediction values.

10. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S1, the performance is controlled by adjusting the filler volume fraction and interfacial modification strength based on the difference between the predicted performance value and the target performance value. Specifically: Calculate the relative error between the predicted performance value and the target performance value, wherein the predicted performance value includes the predicted macroscopic thermal conductivity, the predicted electrical strength, and the predicted flexural strength; A multi-objective optimization function is established based on the relative error, with the optimization objective being to minimize the weighted sum of the relative errors of the predicted macroscopic thermal conductivity, predicted electrical strength, and predicted flexural strength. Key control parameters for macroscopic thermal conductivity, electrical strength and flexural strength are identified. These key control parameters include filler volume fraction and interface modification strength, wherein the interface modification strength is characterized by the functional group density and molecular chain length of the interface modifier. Based on gradient descent or genetic algorithm, the filler volume fraction and interface modification intensity are iteratively optimized within a preset parameter adjustment range, wherein the interface modification intensity is characterized by the functional group density and molecular chain length of the interface modifier. When the value of the multi-objective optimization function converges within the preset tolerance range, or reaches the maximum number of iterations, the optimization process is stopped, and the optimal combination of filler volume fraction and interface modification strength is obtained. Performance is controlled based on the optimal combination of filler volume fraction and interfacial modification strength.

11. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S2, based on the application conditions and expected performance indicators of the high thermal conductivity epoxy resin composite material in dry-type transformers, a calibration sample containing microcracks of a predetermined morphology is prepared. Specifically: Based on the application conditions and expected performance indicators, the preset morphological parameters of microcracks are determined through simulation analysis. The expected performance indicators include thermal conductivity, volume resistivity, flexural strength and heat deformation temperature. The preset morphological parameters include the number of microcracks, the length of microcracks, the width of microcracks, the depth of microcracks and their spatial distribution coordinates. The calibration sample to be prepared is fixed according to its actual assembly orientation in the dry-type transformer winding; By applying controlled mechanical stress or thermal shock treatment, microcracks consistent with the preset morphological parameters are introduced into the fixed calibration sample to obtain the calibration sample containing microcracks.

12. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 11, characterized in that, In step S2, the calibration sample is scanned with multi-frequency electromagnetic waves to acquire and process the electromagnetic wave response signal, generating the electromagnetic wave response characteristic spectrum of the calibration sample. Specifically: Based on the material properties of the calibration sample and the preset morphological parameters of the microcracks, the scanning parameters of the multi-frequency electromagnetic waves are set. The material properties of the calibration samples include thermal conductivity, dielectric constant, and elastic modulus. The scanning parameters include a frequency range of 0.1THz to 2.7THz and a scanning resolution of not less than 6GHz; Based on the scanning parameters, a full-coverage scan is performed on the calibration sample fixed in the actual assembly position of the dry-type transformer winding, and the original electromagnetic wave response signal is acquired in real time. The original electromagnetic wave response signal is preprocessed to obtain a purified signal. The preprocessing includes wavelet threshold noise reduction and matched filtering enhancement. Extract characteristic parameters, including characteristic frequency, characteristic amplitude, reflection coefficient, and projection coefficient, from the purified signal; The characteristic parameters are integrated and normalized to generate the electromagnetic wave response characteristic spectrum of the calibration sample.

13. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 12, characterized in that, The process of integrating and normalizing the characteristic parameters to generate the electromagnetic response characteristic spectrum of the calibration sample includes: Based on the characteristic frequency and the characteristic amplitude, a clustering analysis method is used to perform pattern recognition to obtain a cluster of characteristic parameters representing different microcrack types; Based on the cluster of characteristic parameters, a multi-level micro-defect information superposition algorithm is used to perform spatial domain fusion processing on the reflection coefficient and projection coefficient to obtain enhanced defect spatial distribution characteristics. Based on the cluster of characteristic parameters and the enhanced spatial distribution characteristics of defects, a quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters is constructed. Based on the quantitative mapping relationship, the characteristic parameters of the entire frequency band are normalized and encoded to generate an electromagnetic wave response characteristic spectrum containing characteristic frequency bands, characteristic amplitudes, and spatial distribution coordinates.

14. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 13, characterized in that, The step of performing spatial domain fusion processing on the reflection coefficient and projection coefficient based on the feature parameter cluster and employing a multi-level micro-defect information overlay algorithm to obtain enhanced defect spatial distribution features includes: Based on the cluster of characteristic parameters, the amplitude gradient distribution of the reflection coefficient and the phase delay distribution of the projection coefficient are adaptively divided into multiple data subsets according to the scanning space coordinates. The number of the subsets is positively correlated with the density of the microcrack spatial distribution coordinates. Based on the multiple hierarchical data subsets, the local amplitude maxima of the reflection coefficients and the average phase delay of the projection coefficients within each hierarchical level are extracted to construct the feature vector of the current level. Principal component analysis was used to reduce the dimensionality of the feature vectors at each level and fuse them to obtain the defect probability distribution map corresponding to each level. Based on spatial coordinates, the defect probability distribution maps corresponding to each level are weighted and superimposed to obtain the superimposed defect probability distribution map. The superimposed defect probability distribution map is processed by morphological opening operation to filter out noise and connect adjacent defect regions to obtain enhanced defect spatial distribution features.

15. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S2, based on the electromagnetic wave response characteristic spectrum, a clustering analysis and multi-level micro-defect information superposition algorithm are used to extract the quantitative mapping relationship between microcrack morphology and electromagnetic characteristic parameters, and an electromagnetic feature fingerprint database is established based on the quantitative mapping relationship. Specifically: Based on the electromagnetic wave response characteristic spectrum, the K-means++ clustering algorithm is used to perform unsupervised classification of the characteristic frequency bands and characteristic amplitudes, resulting in an initial mapping relationship lookup table with the cluster center as the index and associated with the mean length and mean width of the corresponding microcracks. Based on the spatial distribution coordinates in the electromagnetic wave response characteristic spectrum, a multi-level micro-defect information superposition algorithm is used to extract the spatial distribution characteristics of defects. Based on the aforementioned spatial distribution characteristics of defects, the correlation features between the depth of microcracks and their spatial distribution coordinates are extracted by Gaussian mixture model clustering. The number of components in the Gaussian mixture model clustering is adaptively determined according to the number of microcracks in the preset morphological parameters. Based on the initial mapping lookup table and the associated features, a microcrack morphology-electromagnetic feature parameter mapping relationship matrix is ​​constructed, which uses feature frequency and feature amplitude as parameters and associates microcrack length, microcrack width, microcrack depth and spatial distribution coordinates. Based on the microcrack morphology-electromagnetic feature parameter mapping matrix, the feature parameters in the electromagnetic wave response feature spectrum are normalized and encoded to establish the electromagnetic feature fingerprint database.

16. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 15, characterized in that, The step of extracting the correlation features between the depth of microcracks and their spatial distribution coordinates using Gaussian mixture model clustering based on the spatial distribution characteristics of the defects includes: The spatial distribution characteristics of the defects are used as the input feature vector for Gaussian mixture model clustering; The number of components for Gaussian mixture model clustering is adaptively determined based on the number of microcracks in the preset morphological parameters. The expectation-maximization algorithm is used to estimate the parameters and iteratively optimize the Gaussian mixture model until the Gaussian mixture model converges. Based on the converged Gaussian mixture model, the posterior probability of each microcrack data point belonging to each Gaussian component is calculated. Based on the maximum a posteriori probability principle, each microcrack data point is assigned to the corresponding Gaussian component to complete defect clustering; Extract the mean vector and covariance matrix of each Gaussian component, where the mean vector represents the typical depth and spatial coordinates of the current type of microcrack, and the covariance matrix represents the distribution pattern of the typical depth and spatial coordinates of the current type of microcrack. Based on the mean vector and covariance matrix of each Gaussian component, a three-dimensional correlation feature matrix between the depth of the microcrack and its spatial distribution coordinates is constructed, which serves as the correlation feature between the depth of the microcrack and its spatial distribution coordinates.

17. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 12, characterized in that, In step S2, during the accelerated aging test simulating the operating conditions of a dry-type transformer, the high thermal conductivity epoxy resin composite material sample undergoes multi-stage electromagnetic wave scanning, and electromagnetic wave scanning data is generated by encoding according to a time series. Specifically: The high thermal conductivity epoxy resin composite material sample was fixed in the aging test device according to the actual assembly orientation of the dry transformer winding. Based on the actual operating conditions of the dry-type transformer, the environmental parameters for the aging test are set, including: a temperature cycling range of -40°C to +120°C, a relative humidity range of 20% to 95%, a mechanical vibration frequency of 10Hz to 200Hz and an amplitude of 0.5g to 5g, and a voltage level of 0.5 times to 1.5 times the rated voltage. Accelerated aging tests are conducted according to a preset aging cycle, which includes multiple consecutive aging stages. Each aging stage includes a heating sub-stage, a heat preservation sub-stage, a cooling sub-stage, and a voltage loading sub-stage. After each aging stage, the high thermal conductivity epoxy resin composite material sample is scanned using the multi-frequency electromagnetic wave scanning parameters to collect the original electromagnetic wave response signal. The collected raw electromagnetic wave response signal is preprocessed to obtain a purified signal. The preprocessing includes wavelet threshold denoising and matched filtering enhancement. From the purified signal, feature parameters including characteristic frequency, characteristic amplitude, reflection coefficient and projection coefficient are extracted to obtain the feature parameters of each stage; According to the aging time series, the characteristic parameters of each stage are encoded in time series to generate electromagnetic wave scanning data.

18. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 17, characterized in that, In step S2, a deep learning model is used to detect and evaluate the electromagnetic feature fingerprint database and electromagnetic wave scanning data to obtain quantitative detection results and quantitative evaluation results, specifically: Construct a deep learning model that includes convolutional neural network branches and long short-term memory network branches; The electromagnetic fingerprint database is input into the branch of the convolutional neural network to extract spatial features and obtain static fingerprint features. The electromagnetic wave scanning data is input into the long short-term memory network branch to extract dynamic temporal features. A cross-modal attention fusion module is used to adaptively weight and fuse static fingerprint features and dynamic temporal features to obtain a fused feature vector; The fused feature vector is input into a fully connected classification layer to classify the defect type and regress the spatial location, and output the type and location of defects in the high thermal conductivity epoxy resin composite sample. The fused feature vectors are simultaneously input into a regression layer to quantitatively predict the defect size and output the size of the defect in the high thermal conductivity epoxy resin composite sample. The dynamic time series characteristics are input into a time series analysis layer to evaluate the aging degree of high thermal conductivity epoxy resin composite material samples and predict the remaining life. The predicted values ​​of aging degree and remaining life of high thermal conductivity epoxy resin composite material samples are output. The quantitative detection results are obtained by integrating the outputs of the classification layer and the regression layer. The quantitative evaluation results are obtained by integrating the output of the time series analysis layer.

19. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 18, characterized in that, In step S2, based on the quantitative detection results and the quantitative assessment results, a health status report for the high thermal conductivity epoxy resin composite material is generated. Specifically, the health status report includes: A planar schematic diagram of a high thermal conductivity epoxy resin composite sample, including the type, location, and size of defects. Aging degree of high thermal conductivity epoxy resin composite material sample as a function of time; Predicted remaining life of high thermal conductivity epoxy resin composite material specimens.

20. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S3, microscopic modeling, mesoscopic modeling, and macroscopic modeling are performed sequentially according to the material design parameters to obtain the formulation and structure of the high thermal conductivity epoxy resin. Specifically: Based on the material design parameters, the composite material component system is pre-designed to determine the basic formulation data, including the epoxy resin matrix type, the compound thermally conductive filler system, the interface modifier and its mass fraction range. Microscale molecular dynamics modeling was performed based on the basic formulation data to obtain microscale interface parameters including interfacial binding energy, interfacial thermal conductivity, matrix glass transition temperature, elastic modulus, and dielectric constant. Based on the microscopic interface parameters, a mesoscopic-scale phase-field method model is used to simulate the evolution of the material structure and obtain mesoscopic structural characteristic parameters including filler dispersion uniformity, interface layer thickness, aggregate size and distribution density. Based on the mesoscopic structural characteristic parameters, macroscopic-scale finite element modeling and multiphysics coupling analysis are performed to predict macroscopic performance parameters including thermal conductivity, bending strength and electrical strength. The macroscopic performance parameters are compared with the preset target performance, and the microscopic interface parameters, mesoscopic structure characteristic parameters and filler volume fraction are adjusted by orthogonal iteration method to perform parameter iterative optimization until the difference from the preset target performance is within the preset range, and the high thermal conductivity epoxy resin formulation and structure are output. The microscale molecular dynamics modeling includes analyzing the glass transition temperature of the matrix, the mesoscale phase field modeling includes controlling the evolution of material structure, and the macroscale finite element modeling includes multiphysics coupling analysis.

21. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 20, characterized in that, In step S3, a high thermal conductivity epoxy resin composite material sample is prepared according to the formula and structure, and terahertz detection is performed. The sample evaluation result is obtained by analyzing the data using a deep learning algorithm. Specifically: A high thermal conductivity epoxy resin composite material sample containing microcracks was prepared according to the aforementioned high thermal conductivity epoxy resin formulation and structure. Wideband terahertz time-domain spectroscopy was performed on the high thermal conductivity epoxy resin composite material sample. By optimizing the photoconductive antenna structure and signal processing algorithm, detection data with a bandwidth of 0.1-2.7THz, a resolution of 6GHz, and a dynamic range of >60dB were obtained. Based on the detection data, a defect fingerprint database containing electromagnetic wave characteristic parameters is constructed, including the reflection coefficient and the transmission coefficient. Based on the aging test data obtained from the defect fingerprint database and orthogonal test method, a mapping relationship between defect features and aging state is established using a deep learning algorithm; The high thermal conductivity epoxy resin composite material sample was tested according to the mapping relationship, and the sample evaluation results were obtained. The sample evaluation results include quantitative detection results and quantitative evaluation results; The quantitative detection results include the type, location, and size of defects in the high thermal conductivity epoxy resin composite material sample; The quantitative assessment results include the aging degree of the high thermal conductivity epoxy resin composite material sample and the predicted value of its remaining life.

22. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 21, characterized in that, In step S3, the material design parameters are corrected based on the sample evaluation results to obtain a material performance dataset. Then, the dry-type transformer structure is designed by configuring the equivalent physical property parameters of the high thermal conductivity epoxy resin coil. Specifically: Based on the quantitative detection results in the sample evaluation results, the filler selection and interface modifier parameters in the high thermal conductivity epoxy resin formulation are optimized by feedback to obtain the corrected material design parameters. Based on the revised material design parameters and the aging degree of the high thermal conductivity epoxy resin composite material specimens in the quantitative evaluation results, a material property dataset including thermal conductivity, electrical strength, and flexural strength is established. Based on the aforementioned material property dataset, the equivalent physical property parameters of a high thermal conductivity epoxy resin coil are configured by studying the thermal conductivity characteristics under different coil winding methods and air passage arrangement methods. Based on the equivalent physical property parameters and the results of the study on the thermal conductivity characteristics of the core and coil, the overall structure of the dry-type transformer is designed, resulting in a dry-type transformer structure that includes the core structure, coil arrangement, and air duct layout.

23. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 22, characterized in that, In step S3, based on the material property dataset, the equivalent physical property parameters of the high thermal conductivity epoxy resin coil are configured by studying the thermal conductivity characteristics under different coil winding methods and air passage arrangements. Specifically: Based on the thermal conductivity in the material property dataset, a porous medium free fluid coupled flow heat transfer model is established, which includes wires, insulating films, high thermal conductivity epoxy resin composite materials, and cooling air channels. Based on the porous medium free fluid coupled flow heat transfer model, and considering the gas flow characteristics under different coil winding methods (layered winding, segmented winding), and different air passage spacing and number arrangement, the gas permeability and heat dispersion coefficient are determined. Based on the gas permeability and heat dispersion coefficient, the thermal conductivity parameters of the high thermal conductivity epoxy resin coil are determined by solving the non-thermal equilibrium flow heat transfer equation and considering the coil temperature distribution characteristics under different winding and arrangement methods. Based on the aforementioned thermal conductivity parameters, and configured according to constraints including coil temperature rise limits, main insulation strength, and inter-turn insulation strength, the equivalent physical property parameters of the high thermal conductivity epoxy resin coil are obtained.

24. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 23, characterized in that, In step S3, based on the dry-type transformer structure and considering the harmonics and power output fluctuations in the renewable energy access scenario, a multiphysics coupled simulation model is constructed and its performance is verified to obtain the performance evaluation results of the dry-type transformer. Specifically: Based on the dry-type transformer structure and the equivalent physical property parameters of the high thermal conductivity epoxy resin coil, a multi-physics field coupling simulation model including electric field, magnetic field, temperature field and stress field is constructed. Based on the harmonic frequency, harmonic content, and power output fluctuation amplitude parameters of the renewable energy access scenario, the boundary conditions and excitation sources of the multiphysics coupling simulation model are set. Based on the boundary conditions and excitation sources of the multiphysics coupling simulation model, the multiphysics coupling simulation model is run and electric field analysis is performed to obtain the electric field intensity distribution and inter-turn voltage distribution. Based on the boundary conditions and excitation source of the multiphysics coupling simulation model, the multiphysics coupling simulation model is run and magnetic field analysis is performed to obtain the magnetic flux density distribution and core loss. Based on the boundary conditions and excitation sources of the multiphysics coupled simulation model, the multiphysics coupled simulation model is run and the temperature field is analyzed to obtain the temperature distribution and hot spot temperature rise. Based on the boundary conditions and excitation sources of the multiphysics coupled simulation model, the multiphysics coupled simulation model is run and stress field analysis is performed to obtain the mechanical stress distribution and deformation displacement. Based on the electric field intensity distribution, inter-turn voltage distribution, magnetic flux density distribution, temperature distribution, hot spot temperature rise, mechanical stress distribution, and deformation displacement, the performance evaluation results of the dry-type transformer are verified.

25. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 24, characterized in that, In step S3, based on the electric field intensity distribution, inter-turn voltage distribution, magnetic flux density distribution, temperature distribution, hot spot temperature rise, mechanical stress distribution, and deformation displacement, the performance evaluation results of the dry-type transformer are verified. Specifically: Based on the electric field strength distribution, check whether the maximum electric field strength is lower than the breakdown field strength threshold of the high thermal conductivity epoxy resin composite material, and obtain the electric field strength check result; Based on the inter-turn voltage distribution, check whether the inter-turn potential difference meets the insulation coordination requirements, and obtain the inter-turn insulation check result; Based on the temperature distribution and hot spot temperature rise, check whether the temperature rise of the highest temperature point is lower than the preset temperature rise limit to obtain the temperature rise performance verification result. Based on the mechanical stress distribution and deformation displacement, check whether the maximum stress value is lower than the yield strength of the high thermal conductivity epoxy resin composite material and whether the deformation is within the allowable range, and obtain the mechanical stability check result; Based on the electric field strength verification results, inter-turn insulation verification results, temperature rise performance verification results, and mechanical stability verification results, a comprehensive evaluation is conducted to obtain the performance assessment results of the dry-type transformer: If all verification results pass, the dry-type transformer is deemed to meet the operating requirements, and the performance evaluation result is deemed qualified. If any verification result fails, the dry-type transformer is deemed not to meet the operating requirements, and the performance evaluation result is deemed unqualified. Based on the failed verification results, the specific substandard performance indicators and their distribution areas in the dry-type transformer structure are identified. The performance indicators include electric field strength, inter-turn insulation, hot spot temperature rise, or mechanical stress.

26. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 25, characterized in that, In step S3, based on the performance evaluation results, with the optimization objectives of maximizing equipment capacity and minimizing volume, and with electric field strength, temperature rise limit, and mechanical stress as constraints, a multi-objective collaborative optimization is performed using the Monte Carlo iterative method and a global optimization algorithm to obtain the optimized global design parameters for the dry-type transformer. Specifically: If the performance evaluation result is qualified, the current dry-type transformer structural parameters will be used as the optimized global design parameters for the dry-type transformer. If the performance evaluation result is unqualified, the structural parameters of the dry-type transformer to be optimized are determined based on the specific unqualified performance indicators and their distribution areas in the dry-type transformer structure. The structural parameters of the dry-type transformer to be optimized include the epoxy resin content of the core, the coil winding method, the air passage layout parameters, the wire diameter and the wire end distance. Based on the structural parameters of the dry-type transformer to be optimized, a multi-objective optimization function is established with the optimization objectives of maximizing equipment capacity and minimizing volume, and with constraints of electric field strength not exceeding the standard, temperature rise not exceeding the limit, and mechanical stress less than the material tolerance value. Based on the multi-objective optimization function, the Monte Carlo iteration method is used to randomly sample within the feasible region of the dry-type transformer structural parameters to be optimized, generating multiple candidate dry-type transformer structural parameter design schemes. According to the global optimization algorithm, the optimization calculation is performed on the multiple candidate dry-type transformer structural parameter design schemes to obtain the candidate dry-type transformer structural parameter design schemes that satisfy all constraints and make the optimization objective optimal, which are used as the optimized global design parameters of the dry-type transformer.

27. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 26, characterized in that, In step S3, a prototype is manufactured according to the optimized global design parameters of the dry-type transformer, and a power frequency-harmonic composite current temperature rise test and a high-frequency oscillation pulse voltage inter-turn insulation test are performed sequentially to obtain a high thermal conductivity epoxy resin dry-type transformer prototype that has been tested and verified. Specifically: Based on the optimized global design parameters of the dry-type transformer, a high thermal conductivity epoxy resin dry-type transformer prototype was manufactured by using high thermal conductivity epoxy resin composite material for coil winding, core assembly and vacuum casting and curing. According to the high thermal conductivity epoxy resin dry molding machine, a power frequency-harmonic composite current temperature rise test was conducted, including: A high-voltage synthetic circuit is constructed by using the harmonic compensation method, which superimposes the 3rd, 5th, and 7th characteristic harmonic currents onto the power frequency current. Pre-embedded thermocouples and infrared thermal imagers were used to measure the temperature rise distribution of the windings, and resistance strain gauges and Raman distributed fiber optic sensors were used to monitor changes in thermal stress. Run continuously until the temperature rise stabilizes, record temperature and stress data, establish the harmonic-temperature rise correspondence and thermal stress distribution model, and obtain the temperature rise and stress assessment results; A high-frequency oscillating pulse voltage inter-turn insulation test was performed using the aforementioned high thermal conductivity epoxy resin dry-type molding machine, including: An oscillating pulse voltage is generated using a high-voltage synthesis circuit with a trigger gap; The partial discharge initiation voltage was measured by boosting the voltage at a rate of 1kV / s, and lightning impulse and switching impulse voltages were applied. By monitoring the discharge signal with a partial discharge detector, the coil was disassembled after the test to observe the carbonization points, an insulation failure early warning threshold was determined, and the inter-turn insulation assessment results were obtained. Based on the temperature rise and stress assessment results and the inter-turn insulation assessment results, a comprehensive determination is made as to whether the high thermal conductivity epoxy resin dry converter prototype passes the test verification: If the test verification is passed, the high thermal conductivity epoxy resin dry deformation sampler shall be regarded as the high thermal conductivity epoxy resin dry deformation sampler verified by the test. If the test verification fails, the specific failure mode of the high thermal conductivity epoxy resin dry-type transformer prototype will be located based on the failure to meet the temperature rise and stress assessment results and / or the failure to meet the inter-turn insulation assessment results, and the global design parameters of the dry-type transformer will be optimized accordingly.

28. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S4, a coil with radial heat dissipation channels is formed by using a high thermal conductivity epoxy resin composite material as the insulating and adhesive medium. Specifically: During the coil winding process, strip-shaped supports are placed at predetermined positions between adjacent coil layers. The strip-shaped supports are made of volatile materials. End baffles are installed at both ends of the wound coil along the axial direction. The inner ring surface of the end baffles is provided with limiting grooves that are adapted to the position and shape of the strip-shaped support. The high thermal conductivity epoxy resin composite material is used to vacuum pressure impregnate the coil that has been wound and has end baffles installed, so that the high thermal conductivity epoxy resin composite material fills the gaps between coils and turns. The coil after vacuum pressure impregnation is subjected to a first-stage curing treatment. The temperature of the first-stage curing treatment is controlled within the range that allows the high thermal conductivity epoxy resin composite material to initially gel but is below the sublimation point of the volatile material. The coil, after the first stage of curing, is subjected to a second stage of gradient temperature curing. The temperature curve of the second stage of gradient temperature curing is set to allow the volatile material to completely sublimate and escape, forming a continuous radial heat dissipation channel between the coil layers, while simultaneously allowing the high thermal conductivity epoxy resin composite material to be completely cured and molded, forming a coil with radial heat dissipation channels.

29. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S4, the stacked iron cores are bonded and cured using a high thermal conductivity epoxy resin composite material to form an integral iron core. Specifically: Multiple stamped iron chips are stacked according to a preset stacking diagram to form a stack body, so that the micro protrusions of adjacent iron chips are staggered and arranged to form a connected resin microchannel mesh inside the stack body. The iron chips have multiple micro protrusions protruding in the stacking direction at preset positions on their edges. In a vacuum environment, a high thermal conductivity epoxy resin composite material preheated to a preset viscosity range is injected and impregnated into the stack, so that the high thermal conductivity epoxy resin composite material fills the gaps between all iron chips through the resin microchannel mesh. Axial pressure is applied to the impregnated stack and gradient temperature is increased for curing. During the gradient temperature increase curing process, the thickness of the insulating layer between the iron chips is controlled and maintained uniformly through the mutual support of the micro bosses. After gradient heating and curing, the integral iron core is obtained through cooling and post-treatment.

30. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 29, characterized in that, The process of applying axial pressure to the impregnated stack and performing gradient temperature curing, wherein the mutual support of the micro-protrusions during the gradient temperature curing process controls and maintains the uniformity of the insulation layer thickness between the iron chips, includes: In a vacuum environment, a preset first axial pressure is applied to the impregnated stack along its stacking direction and maintained for a first preset time, so that excess high thermal conductivity epoxy resin composite material is discharged from the resin microchannel mesh at the edge of the stack. After debinding, the stacked body is transferred to a curing mold, a second axial pressure higher than the first axial pressure is applied, and the stacked body is heated to a first preset temperature at a first heating rate, so that the high thermal conductivity epoxy resin composite material reaches a predetermined viscosity and begins to gel. The temperature is raised to a second preset temperature at a second heating rate and held for a second preset time. The change in the thickness of the stack is monitored in real time by the displacement sensor built into the curing mold, and the second axial pressure is dynamically adjusted based on the change to compensate for the shrinkage of the high thermal conductivity epoxy resin composite material during the curing process and maintain the uniform thickness of the insulation layer between the iron chips.

31. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 30, characterized in that, In step S4, the coil is fitted onto the integral iron core to form an assembly, and an insulating component is installed on the assembly to form an axial heat dissipation channel. Temperature and stress sensors are pre-embedded inside the coil. Specifically: After covering the surface of the core column of the integral iron core with an insulating film, elastic insulating support strips with self-centering function are installed. The elastic insulating support strips are distributed along the core column axis of the integral iron core and arranged in a circumferential array. After the coil is heated to the preset expansion temperature, it is fitted onto the core post of the integral iron core on which the elastic insulating support bar has been installed. After the coil cools and shrinks, the inner wall of the coil is interference-fitted with the elastic insulating support bar and pressed tightly to form the first positioning. Insulating end rings are fitted at both ends of the coil along the axial direction, and U-shaped groove insulating partitions made of high thermal conductivity epoxy resin composite material are inserted between adjacent coils and between the coil and the overall iron core yoke. Each U-shaped groove insulating partition and the elastic insulating support strip are interleaved to form an axial heat dissipation channel extending along the coil axis. Before mounting the coil, the temperature and stress sensors are attached to the preset hot spots and stress concentration areas on the inner wall of the coil. The leads of the temperature and stress sensors are led out along the radial heat dissipation channel and temporarily fixed at the end of the coil through a detachable sealing joint.

32. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, In step S4, the assembly is integrally cast using a high thermal conductivity epoxy resin composite material in a vacuum environment, and then cured under gradient temperature control to form a transformer prototype. Specifically: Before casting, the assembly is vacuum dried and the leads of the temperature and stress sensors are led out through the sealed terminals pre-set on the casting mold. The high thermal conductivity epoxy resin composite material, preheated to the casting viscosity, is injected into the casting mold in two stages under vacuum: In the first stage, the high thermal conductivity epoxy resin composite material is injected from the bottom of the casting mold at the first injection rate, so that the high thermal conductivity epoxy resin composite material preferentially fills the gaps between the coil, the integral iron core and the insulating parts at the bottom of the assembly. In the second stage, the high thermal conductivity epoxy resin composite material is injected from the top of the casting mold at a second injection rate higher than the first injection rate until the assembly is completely submerged in the high thermal conductivity epoxy resin composite material. After the casting is completed, the first negative pressure value is maintained in a vacuum environment and the mixture is left to stand for a first preset time to allow the air bubbles in the high thermal conductivity epoxy resin composite material to escape. Gradient temperature-controlled curing is performed on the casting mold and its internal materials after casting. The temperature of the casting mold and its internal material is raised to the first curing temperature at a first heating rate and kept at that temperature to allow the high thermal conductivity epoxy resin composite material to initially gel. The temperature of the casting mold and the internal material is raised to the second curing temperature at a second heating rate lower than the first heating rate and kept at the temperature to allow the high thermal conductivity epoxy resin composite material to fully cure. The temperature of the casting mold and the internal material is then slowly cooled to the demolding temperature at a controlled cooling rate. During the heat preservation stage of gradient temperature-controlled curing, the heat preservation temperature or heat preservation time is adjusted based on the real-time monitoring of temperature and stress changes of the internal material using temperature and stress sensors led out from the sealed terminal.

33. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 1, characterized in that, The transformer prototype test method described in step S4 includes: An oscillating pulse voltage wave simulating frequent overvoltage conditions is applied to the high-voltage winding of the transformer prototype, while a power frequency voltage is applied to the low-voltage winding. The inter-turn insulation withstand data of the transformer prototype are continuously acquired and recorded at a preset period. A composite current containing the fundamental wave and a specified harmonic is applied to a transformer prototype to simulate load fluctuations caused by renewable energy access. Temperature rise distribution, hot spot data and mechanical vibration spectrum of the transformer prototype are collected and generated by pre-embedded temperature and stress sensors. Based on inter-turn insulation tolerance data, temperature rise distribution, hot spot data, and mechanical vibration spectrum, the thermal conductivity, insulation performance, and mechanical performance of the transformer prototype are evaluated.

34. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 33, characterized in that, The process involves applying oscillating pulse voltage waves simulating frequent overvoltage conditions to the high-voltage winding of the transformer prototype, while simultaneously applying power frequency voltage to the low-voltage winding, and continuously acquiring and recording the inter-turn insulation withstand data of the transformer prototype for a preset period, including: The transformer prototype was placed in a shielded test environment, and the first end of the high voltage winding was connected to the output end of the high voltage pulse generator. The first and last ends of the low voltage winding were short-circuited and connected to one pole of the power frequency voltage source. At the same time, the core and shell of the transformer prototype were grounded. An oscillating pulse voltage wave is applied to the high-voltage winding at a preset repetition frequency and pulse amplitude. The waveform of the oscillating pulse voltage wave is a decaying oscillating waveform, the voltage value of its first peak is a preset multiple of the peak value of the rated phase voltage of the transformer prototype, and its oscillation frequency is within a preset range of the inherent resonant frequency of the transformer prototype winding. While applying an oscillating pulse voltage wave to the high-voltage winding, a power frequency sinusoidal voltage equal to the rated voltage of the transformer prototype is applied to the low-voltage winding, and the phase of the power frequency sinusoidal voltage is controlled so that its peak voltage moment is synchronized with the first peak moment of the oscillating pulse voltage wave. A complete test sequence is formed by continuously applying voltage combinations a preset number of times. In each test sequence, the pulse impact current waveform and partial discharge signal of the high-voltage winding, as well as the power frequency leakage current of the low-voltage winding, are collected synchronously. The transfer function of the winding is calculated based on the pulse impact current waveform, and the inter-turn insulation withstand data of the transformer prototype is analyzed and recorded according to the changes in the partial discharge signal and the power frequency leakage current.

35. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 34, characterized in that, The process involves applying a composite current containing the fundamental wave and a specified harmonic to the transformer prototype to simulate load fluctuations from renewable energy integration. This is achieved by collecting and generating temperature rise distribution, hotspot data, and mechanical vibration spectra of the transformer prototype using pre-embedded temperature and stress sensors. A test current, composed of a power frequency fundamental wave and a specified harmonic with a preset amplitude and phase, is applied to the low-voltage winding of the transformer prototype using a programmable composite current source to simulate a typical harmonic spectrum with a power electronic converter connected. The programmable composite current source is controlled to output the test current according to a preset load cycle curve. In the load cycle curve, the amplitude of the fundamental current changes periodically, and the content of each harmonic is adjusted according to a preset ratio to simulate the load fluctuation of renewable energy access. During the application of the test current, data from temperature and stress sensors embedded inside the coil are collected simultaneously, and vibration acceleration signals are collected at preset monitoring points on the outer shell of the transformer prototype. The collected temperature data is mapped and fitted to the three-dimensional structural model of the transformer prototype to generate a dynamic temperature rise distribution map during the operation of the transformer prototype, and the temperature area exceeding the preset threshold is identified as hot spot data. Time-frequency analysis was performed on the collected stress data and vibration acceleration signals to extract characteristic frequency components related to the specified harmonic frequency, and the mechanical vibration spectrum of the transformer prototype under harmonic excitation was synthesized.

36. The comprehensive preparation and performance assurance method for a high thermal conductivity epoxy resin-based dry-type transformer according to claim 35, characterized in that, The thermal conductivity, insulation performance, and mechanical performance of the transformer prototype are evaluated based on inter-turn insulation tolerance data, temperature rise distribution, hotspot data, and mechanical vibration spectrum, including: Based on the dynamic temperature rise distribution map and the hot spot data, the thermal conductivity of the transformer prototype is evaluated using computer-equivalent thermal network parameters. Based on the inter-turn insulation tolerance data, a quantitative index of insulation aging degree is calculated to evaluate the insulation performance of the transformer prototype. The vibration spectrum of the structure is compared with the theoretical spectrum obtained by modal simulation based on the three-dimensional model of the transformer prototype to evaluate the mechanical performance of the transformer prototype.

37. A comprehensive preparation and performance assurance system for a high thermal conductivity epoxy resin-based dry-type transformer, applied to the comprehensive preparation and performance assurance method for the high thermal conductivity epoxy resin-based dry-type transformer according to any one of claims 1-36, characterized in that, include: The performance control module is used to control the performance of the high thermal conductivity epoxy resin composite material according to the preset material performance target value, thereby obtaining optimized material design parameters. The material design parameters include epoxy resin matrix parameters, compounded thermally conductive filler parameters, and interface modifier parameters. The detection and evaluation module is used to detect and evaluate high thermal conductivity epoxy resin composite material samples prepared according to the material design parameters using a detection and evaluation method based on high thermal conductivity epoxy resin composite materials, obtain sample evaluation results, and correct the material design parameters based on the sample evaluation results to obtain the corrected material design parameters. The design and verification module is used to perform multi-scale modeling, structural design, multi-physics field coupled simulation verification and multi-objective collaborative optimization of dry-type transformers based on the modified material design parameters, using a high thermal conductivity epoxy resin dry-type transformer design and verification method, to obtain the optimized global design parameters of the dry-type transformer. The prototype manufacturing module is used to manufacture a high thermal conductivity epoxy resin-based dry-type transformer prototype according to the optimized global design parameters of the dry-type transformer using the transformer prototype manufacturing method, and to conduct comprehensive performance tests and verifications on the transformer prototype using the transformer prototype testing method, thereby completing the comprehensive preparation and performance assurance of the high thermal conductivity epoxy resin-based dry-type transformer. The performance control method is used to determine the filler combination, interface modifier, and related interface and structural parameters of the high thermal conductivity epoxy resin composite material, so as to achieve precise control of the macroscopic thermal conductivity, electrical strength, and flexural strength of the material; the detection and evaluation method is used to quantitatively detect the defect type, location, and size of the composite material sample, and evaluate the aging degree and remaining life of the sample; the dry-type transformer design and verification method is used to complete the microscopic, mesoscopic, and macroscopic multi-scale modeling of the dry-type transformer, as well as the structural design and multi-physics field coupling simulation verification of the dry-type transformer based on the renewable energy access scenario; the transformer prototype manufacturing method is used to complete the molding and overall casting and curing of the transformer coil and core; and the transformer prototype testing method is used to verify the inter-turn insulation, thermal conductivity, and mechanical properties of the transformer prototype.

38. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the comprehensive preparation and performance assurance method for the high thermal conductivity epoxy resin-based dry-type transformer as described in any one of claims 1-36.

Images