Epoxy insulating material aging characteristic joint evaluation characterization method

Abstract

The invention belongs to the technical field of epoxy material aging evaluation methods, particularly relates to a joint evaluation characterization method for aging characteristics of an epoxy insulating material, and aims to solve the problems that in the prior art, a method for simultaneously carrying out comprehensive evaluation on aging of epoxy resin on thermal, electromagnetic and mechanical property changes is lacked, and most of existing methods adopt simple threshold comparison. And the accuracy and objectivity of an evaluation result are limited. The characterization method comprises the following steps: providing an epoxy insulating material sample; comprehensive aging treatment of electrical aging, thermal aging and mechanical aging is applied to the epoxy insulating material sample so as to obtain epoxy insulating material samples aged at different time lengths; respectively measuring a thermal performance parameter, an electromagnetic performance parameter and a mechanical performance parameter corresponding to each aging time length; and inputting the obtained performance parameters into the neural network model, and outputting the aging grade and the aging factor of the epoxy insulating material so as to complete the joint evaluation of the aging characteristics of the epoxy insulating material.

Description

Technical Field

[0001] The present invention belongs to the technical field of epoxy material aging evaluation methods, and particularly relates to a combined evaluation and characterization method for the aging characteristics of epoxy insulating materials. Background Art

[0002] Epoxy resin-based insulating materials have been widely used in various electrical equipment such as transformers, insulators, and cable joints due to their excellent electrical insulation performance and mechanical strength. The insulation performance of these epoxy insulating materials is crucial for the safe and reliable operation of electrical equipment. However, as the service time of the equipment increases, epoxy insulating materials will inevitably be gradually aged under the combined influence of environmental factors and working stresses. Factors such as long-term electric field action, periodic high and low temperature changes, humidity erosion, and mechanical stress vibration will all promote physical and chemical structure changes in epoxy materials, such as molecular chain breakage, change in crosslinking density, formation of microcracks and voids, etc. These cumulative aging effects will lead to deterioration of the dielectric properties, thermal conductivity, and mechanical properties of the material, thereby weakening the overall performance of the electrical equipment and posing potential safety hazards. Therefore, accurate evaluation of the aging characteristics of epoxy insulating materials is of great significance for the stable operation of the power system.

[0003] Currently, the evaluation of the aging degree of epoxy insulating materials mainly relies on testing methods in a single field or a simple combination of a few means. Traditionally, infrared thermal imaging technology is often used to study the thermal properties of materials, such as finding hot spots or estimating changes in the thermal conductivity of materials by observing the surface temperature distribution of energized equipment; microwave dielectric testing technology is used to characterize the electromagnetic properties of materials, such as evaluating the dielectric aging degree of materials by measuring the dielectric constant and dielectric loss factor; and ultrasonic nondestructive testing technology is often used to evaluate the mechanical properties and internal defect conditions of materials, such as judging changes in the elastic modulus of materials or internal cracks by the propagation speed and attenuation of sound waves in the material. The above detection technologies provide valuable information in their respective fields, but the information dimensions obtained by each method are limited and can only reflect the characteristic changes of epoxy materials after aging from a single angle. Since the aging of epoxy insulating materials is the result of the combined action of multiple mechanisms, and different aging mechanisms (electrical aging, thermal aging, environmental aging, mechanical stress aging, etc.) have different effects on the material properties, a single detection method is difficult to comprehensively and accurately characterize the overall aging state of the material. For example, the surface temperature change obtained only by infrared thermal imaging cannot reflect the deterioration of the internal dielectric properties of the material, and simple measurement of dielectric parameters cannot reveal the decrease in the mechanical strength of the material. Various traditional detection methods are difficult to take into account the whole picture of the aging characteristics of materials.

[0004] Existing methods for evaluating the aging of epoxy insulation materials still have technical limitations and pain points. First, there is a lack of a comprehensive evaluation method that simultaneously considers changes in thermal, electromagnetic, and mechanical properties. Currently, few literatures or technologies effectively integrate and analyze the detection results across different fields such as infrared, microwave, and ultrasonic. Second, in terms of the analysis of aging data, most methods use empirical judgment or simple threshold comparison, unable to fully explore the correlation laws behind multi-parameter changes, and the accuracy and objectivity of the evaluation results are limited. In addition, the data formats obtained by different detection technologies are diverse and the dimensions are different, making it difficult for manual comprehensive analysis, and subjective factors may affect the judgment, which to a certain extent hinders the standardization of aging evaluation. In summary, the existing technologies are difficult to meet the need for a comprehensive and accurate evaluation of the aging characteristics of epoxy insulation materials, and there is an urgent need for a new technical solution to overcome the above bottlenecks. Summary of the Invention

[0005] The purpose of the present invention is to solve the above deficiencies and provide a method for jointly evaluating and characterizing the aging characteristics of epoxy insulation materials.

[0006] A method for jointly evaluating and characterizing the aging characteristics of epoxy insulation materials includes the following steps: Provide a sample of epoxy insulation material; Apply a comprehensive aging treatment of electrical aging, thermal aging, and mechanical aging to the epoxy insulation material sample to obtain epoxy insulation material samples after aging for different durations; Measure the thermal performance parameters, electromagnetic performance parameters, and mechanical performance parameters corresponding to each aging duration respectively; Input the obtained performance parameters into a neural network model, and output the aging grade and aging factor of the epoxy insulation material to complete the joint evaluation of the aging characteristics of the epoxy insulation material.

[0007] Further, the aging durations include 0h, 24h, 48h, 96h, 192h, 384h, and 768h.

[0008] Further, the thermal performance parameters at least include the planar thermal conductivity; the steps for obtaining the planar thermal conductivity are: Obtain the surface temperature distribution image of the epoxy insulation material sample after aging for different durations; Perform curve fitting on the central region of the temperature distribution, and calculate the planar thermal conductivity of the sample at the corresponding aging duration.

[0009] Further, the electromagnetic performance parameters at least include the relative permittivity; the steps for obtaining the relative permittivity are: Use a cavity resonator to measure the resonant frequency and Q factor of the empty cavity respectively; Place the epoxy insulation material samples after aging for different durations into the cavity and measure the resonant frequency and Q factor; The relative permittivity of the specimen at the corresponding aging duration is calculated based on the resonance frequency shift.

[0010] Furthermore, the electromagnetic property parameters further include complex permittivity; the steps for obtaining the complex permittivity are as follows: Construct a microwave transmission / reflection test system using a vector network analyzer; Place the epoxy insulation material specimens aged for different durations in the transmission / reflection device, and measure the curves of the transmission coefficient and reflection coefficient varying with frequency; Invert the complex permittivity of the specimen at each aging duration based on the variation curves.

[0011] Furthermore, the electromagnetic property parameters also include conductivity; the steps for obtaining the conductivity are as follows: Extract the dielectric loss corresponding to the imaginary part of the complex permittivity; calculate the conductivity of the epoxy insulation material specimens at each aging duration to characterize the change of conductivity with aging.

[0012] Furthermore, the mechanical property parameters at least include elastic modulus and Poisson's ratio; the steps for obtaining the elastic modulus and Poisson's ratio are as follows: Conduct tensile tests on dumbbell-shaped specimens of epoxy insulation materials aged for different durations; Calculate the elastic modulus based on the stress–strain curve, and calculate Poisson's ratio through the ratio of longitudinal and transverse strains.

[0013] Furthermore, the mechanical property parameters further include ultrasonic sound velocity and density; the steps for obtaining the ultrasonic sound velocity are as follows: Measure the length of the specimen along the ultrasonic propagation direction; Couple the ultrasonic longitudinal wave probe of the ultrasonic sound velocity measuring device to one end of the specimen and emit a pulse signal, and record the time difference between the first wave and the reflected wave; Calculate the ultrasonic sound velocity, and evaluate the change of the mechanical integrity of the specimen material in combination with the density.

[0014] Furthermore, the ultrasonic sound velocity measuring device includes an ultrasonic pulse generator, the ultrasonic longitudinal wave probe with a working frequency of 2 MHz to 5 MHz, an ultrasonic couplant, an oscilloscope with a gain of 15 dB to 25 dB, and a computer for data processing.

[0015] Furthermore, the neural network model is a multi-layer forward neural network model, including an input layer, at least two hidden layers, and two output channels; the input layer receives thermal property parameters, electromagnetic property parameters, and mechanical property parameters; the two output channels respectively output the aging grade classification result and the aging factor regression result of the epoxy insulation material.

[0016] Advantages of the present invention: A combined evaluation and characterization method for the aging characteristics of epoxy insulation materials provided by the present invention simultaneously introduces three complementary information chains of thermal performance parameters, electromagnetic performance parameters, and mechanical performance parameters in the same evaluation framework, collects and quantifies the aging characteristics of epoxy insulation materials in the three key performance dimensions of heat, dielectric, and mechanics at one time, and overcomes the limitation that existing evaluation means rely on a single physical field and cannot comprehensively reflect complex aging mechanisms. Since the electro-thermal-mechanical comprehensive aging treatment can simulate the multi-field coupling conditions in the actual service process of equipment, the generated data is more representative of engineering; and after the multi-source parameters are uniformly normalized and input into the multi-layer forward neural network, the network can automatically mine non-linear correlations and output discrete aging levels and continuous aging factors to achieve refined prediction of the remaining life of the material. This method can obtain high-value indicators such as thermal conductivity, dielectric constant, and elastic modulus without damaging the specimen, significantly improving the test safety and repeatability; at the same time, the neural network inference only takes hundreds of milliseconds, which can meet the on-site rapid diagnosis requirements. Generally speaking, this method not only greatly improves the accuracy and resolution of the aging state determination of epoxy insulation materials, but also provides a quantifiable, portable, and extensible technical basis for the condition-based maintenance, life management, and risk warning of power equipment, and has significant technical effects of reducing maintenance costs and ensuring the long-term safe operation of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flowchart of a combined evaluation and characterization method for the aging characteristics of epoxy insulation materials provided for a specific embodiment.

[0018] Figure 2 A schematic diagram of a neural network model provided for a specific embodiment.

[0019] Figure 3 An aging factor loss function provided for a specific embodiment.

[0020] Figure 4 An aging level loss function provided for a specific embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] The following further specifically describes a combined evaluation and characterization method for the aging characteristics of epoxy insulation materials described in the present invention in conjunction with embodiments. For the sake of simplicity of description, this document cannot enumerate all alternative technical features and implementation schemes included in the present invention. Therefore, those skilled in the art should be aware that any technical feature and implementation scheme in this embodiment do not limit the protection scope of the present invention, and this protection scope includes any alternative technical features and implementation schemes that those skilled in the art can obtain without creative labor. Specifically, any implementation scheme obtained by replacing any technical feature in the present invention or combining any two or more technical features provided by the present invention should be within the protection scope of the present invention.

[0022] This embodiment provides a combined evaluation and characterization method for the aging characteristics of epoxy insulation materials, including the following steps: Provide epoxy insulation material specimens; Apply comprehensive aging treatments of electrical aging, thermal aging, and mechanical aging to the epoxy insulation material specimens to obtain epoxy insulation material specimens after aging for different durations; Measure the thermal performance parameters, electromagnetic performance parameters, and mechanical performance parameters corresponding to each aging duration respectively; Input the obtained performance parameters into a neural network model, and output the aging grade and aging factor of the epoxy insulation material to complete the combined evaluation of the aging characteristics of the epoxy insulation material.

[0023] In a specific embodiment, as Figure 1 shown. First, prepare an epoxy resin matrix according to the standards of GB / T1040-2006 and ISO527-2, and use a two-stage process of room temperature and post-curing to obtain dumbbell-shaped mechanical specimens with dimensions of 100mm×10mm×4mm and cuboid electromagnetic-thermal specimens with dimensions of 20mm×20mm×4mm. Subsequently, place the specimens in a comprehensive aging test chamber, which can simultaneously apply a constant temperature cycling thermal field of 180°C, a DC electric field of 10kV / mm, and an alternating compressive-tensile mechanical load of ±5MPa, so that the multi-fields of electricity-thermal-mechanics act on the surface and inside of the material simultaneously. After taking out the specimens at preset time nodes, perform infrared thermal imaging, microwave dielectric testing, and tensile-ultrasonic integrated testing respectively to obtain complete thermal, electromagnetic, and mechanical performance parameters of the three categories. Finally, input all the parameters into a pre-trained forward neural network model, and the model outputs the aging grade (1-5 levels) and a continuous aging factor of 0-1 in real time, realizing the combined evaluation of the entire life cycle of the epoxy insulation material.

[0024] In some embodiments, the aging durations include 0h, 24h, 48h, 96h, 192h, 384h, and 768h.

[0025] In this embodiment, aiming at the time scale of aging kinetics, the sampling time points are set to 0h, 24h, 48h, 96h, 192h, 384h, and 768h, a total of seven levels of accelerated aging durations. Aging is stopped and cooling is carried out every time the specified duration is reached. After the specimens are stabilized at room temperature for 1h, subsequent detections are carried out. By comparing the data of each time period, the performance evolution trajectories in the initial, middle, and late stages of aging can be intuitively presented, providing clear time series labels for the neural network and improving the discrimination accuracy of modeling.

[0026] In some embodiments, the thermal performance parameters at least include the planar thermal conductivity; the steps for obtaining the planar thermal conductivity are: Obtain the surface temperature distribution images of the epoxy insulation material specimens after aging for different durations; Perform curve fitting on the central region of the temperature distribution, and calculate the planar thermal conductivity of the specimen at the corresponding aging duration.

[0027] In a specific embodiment, the steady-state infrared thermal imaging method is used to obtain the thermal properties. Place the cuboid specimen in the center of the perforated platform of the aluminum dark box, and uniformly heat the bottom through the hole with a 300W halogen lamp to form a radial temperature gradient on the upper surface of the specimen. Use an InSb cooled infrared camera with a sensitivity of 15mK and a spatial resolution of 1280×1024px to take the temperature distribution after reaching a steady state. Select the data in the area with a hole center radius of 5mm and fit it with a parabola model. The parabola model is:

[0028] where a, b, c are undetermined parameters.

[0029] Use the least squares method to fit the collected data to the parabola model and calculate the undetermined parameters.

[0030] Calculate the temperature curvature through the following formula:

[0031] where, at the hole center ( r=0 ), the curvature simplifies to: This is because at the center point, the first derivative term b will disappear.

[0032] Calculate the planar thermal conductivity based on the temperature curvature to evaluate the attenuation of the heat conduction ability. The formula derivation is as follows:

[0033]

[0034] where, is the temperature at a distance r from the hole center, is the absorbed power density, is the internal thermal conductivity of the epoxy material, d is the sample thickness, C is a constant.

[0035] In some embodiments, the electromagnetic property parameters at least include the relative permittivity; the steps for obtaining the relative permittivity are: Use a cavity resonator to measure the resonant frequency and Q factor of the empty cavity respectively; Place the epoxy insulation material specimens aged for different durations into the cavity and measure the resonant frequency and Q factor; The relative permittivity of the specimen at the corresponding aging duration is calculated based on the resonance frequency shift.

[0036] In a specific embodiment, the dielectric properties of the material are utilized to affect the resonance frequency and Q factor of the resonator, and the dielectric constant is measured by the dielectric resonator method. The formula is as follows:

[0037]

[0038] Where, is the resonance frequency, c is the speed of light, is the relative permittivity of the sample, p and m are specific values related to the resonance mode, a is the radius of the resonator, and d is the height of the resonator.

[0039] When aging causes an increase in polar groups or the initiation of microcracks, the dielectric constant increases accordingly, which can reflect the change in dielectric energy storage ability.

[0040] In some embodiments, the electromagnetic property parameter further includes the complex permittivity; the steps for obtaining the complex permittivity are as follows: Construct a microwave transmission / reflection test system using a vector network analyzer; Place the epoxy insulation material specimens aged for different durations in the transmission / reflection device, and measure the curves of the transmission coefficient and reflection coefficient varying with frequency; Invert the complex permittivity of the specimen at each aging duration based on the variation curves.

[0041] In a specific embodiment, the microwave transmission / reflection method infers the dielectric properties of the material by measuring the transmission (Transmission) and reflection (Reflection) coefficients when the microwave signal passes through the material sample. The relevant formula is as follows:

[0042] Where, is the transmission coefficient and is the reflection coefficient.

[0043] In some embodiments, the electromagnetic property parameter further includes the conductivity; the steps for obtaining the conductivity are as follows: Extract the dielectric loss corresponding to the imaginary part of the complex permittivity; calculate the conductivity of the epoxy insulation material specimen at each aging duration to characterize the change in conductivity with aging.

[0044] In a specific embodiment, the conductivity of the aged epoxy material is measured using the dielectric loss method. At microwave frequencies, the dielectric loss of the material is related to its conductivity. The imaginary part of the complex dielectric constant reflects the absorption ability of the material, from which the conductivity can be inferred. The specific formula is as follows:

[0045] Where, σ is the conductivity to be solved, ω is the angular frequency, and the calculation formula is ω =2 πf , f is the frequency of the signal, is the permittivity in vacuum, is the imaginary part of the complex dielectric constant, representing the dielectric loss.

[0046] When aging causes moisture absorption or crack penetration, the σ value increases exponentially and the energy loss increases significantly. By comparing the increments of σ for different aging durations, the evolution of the conduction path of the insulating material can be quantitatively described, providing a precursor criterion for the risk of dielectric breakdown in the electric field.

[0047] In some embodiments, the mechanical property parameters at least include the elastic modulus and Poisson's ratio; the steps for obtaining the elastic modulus and Poisson's ratio are as follows: Conduct a tensile test on dumbbell-shaped specimens of epoxy insulating materials aged for different durations; Calculate the elastic modulus based on the stress–strain curve, and calculate Poisson's ratio through the ratio of longitudinal to transverse strain.

[0048] In a specific embodiment, an Instron5969 micro-controlled electronic universal testing machine is used for the mechanical property test, and the loading rate is 2 mm / min. Biaxial strain gauges are pasted at the clamping ends of the dumbbell specimens to record the longitudinal and transverse strains in real time. The initial elastic modulus E can be obtained from the slope of the stress-strain curve. With the action of alternating mechanical loads and thermal-oxidation, the fracture of molecular chain segments and the change of crosslinking density cause E to decrease, characterizing the attenuation of the material stiffness and transverse shrinkage.

[0049] In some embodiments, the mechanical property parameters further include the ultrasonic sound velocity and density; the steps for obtaining the ultrasonic sound velocity are as follows: Measure the length of the specimen along the ultrasonic propagation direction; Couple the ultrasonic longitudinal wave probe of the ultrasonic sound velocity measuring device to one end of the specimen and emit a pulse signal, and record the time difference between the first wave and the reflected wave; Calculate the ultrasonic sound velocity, and evaluate the change in the mechanical integrity of the specimen material in combination with the density.

[0050] In some specific embodiments, ultrasonic technology is used to evaluate the mechanical properties of materials. According to the relationship between the sound velocity and the mechanical constitutive properties of the materials, the ultrasonic sound velocity is measured to evaluate the mechanical properties of the aged epoxy resin. Specifically, the calculation formulas for the ultrasonic sound velocity and density are as follows:

[0051]

[0052] Among them, is the longitudinal wave sound velocity of the epoxy insulation material; ρ is the density of the epoxy insulation material; E is the elastic modulus of the epoxy insulation material; v is the Poisson's ratio of the epoxy insulation material.

[0053] The ultrasonic sound velocity can be obtained by dividing the length of the cubic epoxy specimen by the time difference between the initial wave and the reflected wave. The formula is as follows:

[0054] Among them, c is the velocity of the ultrasonic wave, L is the length of the cubic epoxy specimen, t is the time difference between the arrival of the initial wave and the reflected wave.

[0055] In some embodiments, the ultrasonic sound velocity measuring device includes an ultrasonic pulse generator, an ultrasonic longitudinal wave probe with a working frequency of 2 MHz to 5 MHz, an ultrasonic coupling agent, an oscilloscope with a gain of 15 dB to 25 dB, and a computer for data processing.

[0056] In a specific embodiment, all channels of the ultrasonic system are calibrated with a pulse width of 500 ns and a -6 dB bandwidth. The probe frequency can be switched between 2 MHz and 5 MHz to adapt to specimens with different attenuation levels; the gain window of the oscilloscope is 15 - 25 dB. A lower gain can be used in the early aging stage to avoid saturation, and the gain can be increased in the late aging stage with high scattering to maintain the signal-to-noise ratio, realizing seamless monitoring of the full-cycle specimens.

[0057] In some embodiments, the neural network model is a multi-layer forward neural network model, including an input layer, at least two hidden layers, and two output channels; the input layer receives thermal performance parameters, electromagnetic performance parameters, and mechanical performance parameters; the two output channels respectively output the aging grade classification result and the aging factor regression result of the epoxy insulation material.

[0058] In a specific embodiment, the epoxy resin aging characteristic evaluation model based on a neural network has a network structure including an input layer, two hidden layers, and two output layers. The input layer is used to receive 10 characteristic parameters, including: thermal conductivity, temperature gradient, thermal diffusivity, dielectric constant, dielectric loss, reflection coefficient, tensile strength, elastic modulus, ultrasonic propagation velocity, and aging time. The first hidden layer after the input layer includes 6 neurons, and the second hidden layer includes 3 neurons. Both hidden layers use the ReLU activation function (Rectified Linear Unit). The neurons in the two hidden layers are fully connected to ensure full information propagation and feature extraction. The output layer consists of two sub-networks: The first output layer (classification task): It is used to predict the aging level (levels 1-5), includes 5 nodes, uses the Softmax activation function, and is optimized using the cross-entropy loss function (Cross-Entropy Loss).

[0059] The second output layer (regression task): It is used to predict the aging factor (a continuous value between 0 and 1), includes 1 node, uses the Sigmoid activation function, and uses the mean squared error (Mean Squared Error, MSE) as the loss function.

[0060] The schematic diagram of the neural network model structure is as Figure 2 shown. Figure 2 It reveals the neural network modeling structure and information flow for aging characteristic evaluation: The input layer receives multi-dimensional characteristic parameters obtained from infrared, microwave, and ultrasonic detections. After feature extraction through several hidden layers, the output layer is divided into two branches, respectively outputting the aging level and the aging factor. Through Figure 2 It can be seen that this neural network model fuses the data obtained by different detection means and constructs a non-linear mapping relationship between the aging state and multiple parameters. Using the neural network model can fuse and analyze multi-source information of thermology, electromagnetics, and mechanics, so as to realize the intelligent discrimination and prediction of the aging degree of epoxy insulating materials.

[0061] More specifically, the training and optimization of this neural network model include the following steps: First, the dataset is divided into a training set, a validation set, and a test set in the ratio of 8:1:1 to ensure the generalization performance of the model. The input characteristic parameters are all standardized (Standardization) before being input into the model, so that each feature value is transformed into a distribution with a mean of 0 and a standard deviation of 1, in order to improve the stability and convergence speed of model training. During the training process, the Adam optimizer is used for parameter update, and its initial learning rate is 0.001. The optimization algorithm is based on Adaptive Moment Estimation, which can dynamically adjust the learning rate, thereby improving the training efficiency and accelerating convergence. In terms of the design of the loss function, for the dual-output task of the model, a weighted loss strategy for multi-task learning is adopted. Specifically, the loss function for the classification task is Cross-Entropy, which is used to predict the aging level (levels 1-5); the loss function for the regression task is Mean Squared Error (MSE), which is used to predict the aging factor (a continuous value between 0 and 1).

[0062] The graph of the aging factor loss function is as Figure 3 shown. Figure 3 It reveals the loss function design or training convergence of the aging factor regression task in the neural network model. Generally, this graph shows the change trend of the Mean Squared Error (MSE) loss of the aging factor during the training process with the number of iterations in the form of a curve. By Figure 3 it can be seen that the loss value is relatively high at the beginning of training and gradually decreases and stabilizes as the training iterations progress, indicating that the model is learning this continuous indicator of the aging factor. The modeling logic reflected in this appendix is to optimize the network using an appropriate regression loss function, and the key conclusion is that the model can successfully learn and output a continuous value (aging factor) within the range of 0 to 1 that reflects the aging degree. Figure 3 The modeling logic reflected in this appendix is to optimize the network using an appropriate regression loss function, and the key conclusion is that the model can successfully learn and output a continuous value (aging factor) within the range of 0 to 1 that reflects the aging degree.

[0063] The graph of the aging level loss function is as Figure 4 shown. Figure 4 Corresponding to the loss function of the aging level classification task in the neural network model, Cross-Entropy is usually adopted. This graph shows the change of the classification loss during the training process or the mathematical form of the loss function. By Figure 4 it can be seen that as the training progresses, the loss of the classification task gradually decreases, indicating that the prediction accuracy of the model for the aging level (discrete levels 1-5) is improving. The key points of the modeling structure revealed by this graph are to introduce the classification loss to guide the network to learn the discrimination of discrete aging levels, and the model can converge in the multi-classification task and correctly classify the specimens into the corresponding aging levels.

[0064] For those of ordinary skill in the art, other various forms of changes or modifications can be made on the basis of the above description. It is not necessary and impossible to exhaustively list all the embodiments here, and the obvious changes or modifications derived therefrom still fall within the protection scope of the claims of this invention.

Claims

1. A combined evaluation and characterization method for the aging characteristics of epoxy insulation materials, characterized in that, It includes the following steps: Provide epoxy insulation material specimens; Apply comprehensive aging treatment of electrical aging, thermal aging and mechanical aging to the epoxy insulation material specimens to obtain epoxy insulation material specimens after aging for different durations; Measure the thermal performance parameters, electromagnetic performance parameters and mechanical performance parameters corresponding to each aging duration respectively; Input the obtained performance parameters into a neural network model, and output the aging grade and aging factor of the epoxy insulation material to complete the joint evaluation of the aging characteristics of the epoxy insulation material.

2. The method according to claim 1, wherein The aging durations include 0h, 24h, 48h, 96h, 192h, 384h and 768h.

3. The method according to claim 1, wherein The thermal performance parameters at least include planar thermal conductivity; the steps for obtaining the planar thermal conductivity are: Obtain the surface temperature distribution images of the epoxy insulation material specimens after aging for different durations; Perform curve fitting on the central region of the temperature distribution, and calculate the planar thermal conductivity of the specimen at the corresponding aging duration.

4. The method according to claim 1, wherein The electromagnetic performance parameters at least include relative permittivity; the steps for obtaining the relative permittivity are: Measure the resonant frequency and Q factor of the empty cavity using a cavity resonator respectively; Measure the resonant frequency and Q factor after placing the epoxy insulation material specimens after aging for different durations into the cavity; Calculate the relative permittivity of the specimen at the corresponding aging duration based on the resonant frequency shift.

5. The method according to claim 4, characterized in that, The electromagnetic performance parameters further include complex permittivity; the steps for obtaining the complex permittivity are: Construct a microwave transmission / reflection test system using a vector network analyzer; Place the epoxy insulation material specimens after aging for different durations in the transmission / reflection device, and measure the curves of the transmission coefficient and reflection coefficient varying with frequency; Invert the complex permittivity of the specimen at each aging duration based on the variation curves.

6. The method according to claim 5, wherein The electromagnetic performance parameters also include conductivity; the steps for obtaining the conductivity are: Extract the dielectric loss corresponding to the imaginary part of the complex permittivity; calculate the conductivity of the epoxy insulation material specimens at each aging duration to characterize the change of conductivity with aging.

7. The method according to claim 1, characterized in that, The mechanical performance parameters at least include elastic modulus and Poisson's ratio; the steps for obtaining the elastic modulus and Poisson's ratio are: Conduct tensile tests on dumbbell-shaped specimens of epoxy insulation materials after aging for different durations; Calculate the elastic modulus based on the stress-strain curve, and calculate Poisson's ratio through the ratio of longitudinal and transverse strains.

8. The method according to claim 7, characterized in that The mechanical performance parameters further include ultrasonic sound velocity and density; the steps for obtaining the ultrasonic sound velocity are: Measure the length of the specimen along the ultrasonic propagation direction; Couple the ultrasonic longitudinal wave probe of the ultrasonic sound velocity measuring device to one end of the specimen and emit a pulse signal, and record the time difference between the first wave and the reflected wave; Calculate the ultrasonic sound velocity, and evaluate the change of the mechanical integrity of the specimen material in combination with the density.

9. The method according to claim 8, wherein The ultrasonic sound velocity measuring device includes an ultrasonic pulse generator, the ultrasonic longitudinal wave probe with a working frequency of 2 MHz to 5 MHz, ultrasonic coupling agent, an oscilloscope with a gain of 15 dB to 25 dB, and a computer for data processing.

10. The method according to claim 1, characterized in that, The neural network model is a multi-layer feedforward neural network model, including an input layer, at least two hidden layers, and two output channels; the input layer receives thermal performance parameters, electromagnetic performance parameters, and mechanical performance parameters; the two output channels respectively output the aging grade classification result and the aging factor regression result of the epoxy insulating material.

Images