Method for detecting influence of temperature and mechanical deformation accumulation on short-circuit resistance of commutation inductance

Abstract

The application provides a temperature and mechanical deformation cumulative effect detection method for short-circuit resistance of a converter transformer, belongs to the technical field of short-circuit resistance of a converter transformer, and is characterized in that, through precise experiments and data analysis, the mechanical response characteristics of the converter transformer under different temperatures and dynamic force load conditions are deeply researched. First, material characteristic parameters are acquired, and temperature and mechanical load experiments are conducted on a dynamic mechanics tension and pressure machine; advanced equipment such as a three-dimensional laser scanner, a temperature sensor, a displacement sensor and a high-speed camera system are used to record the temperature field distribution, mechanical response, residual deformation and insulation distance change of the components of the converter transformer in detail. Through finite element analysis and data processing, a dynamic characteristic database under the cumulative action of temperature and mechanical deformation is established, and a correlation model is trained by using machine learning technology, so that the problem that the prior art lacks a detection method capable of comprehensively evaluating the interaction of temperature-mechanical deformation-electrical performance is solved.

Description

Technical Field

[0001] This invention belongs to the field of converter transformer short-circuit withstand technology, specifically, it relates to a method for detecting the influence of temperature and mechanical deformation accumulation on the short-circuit withstand performance of converter transformers. Background Technology

[0002] As a key piece of equipment in the power system, the operational reliability of converter transformers directly affects the safety and stability of the entire power grid. During actual service, converter transformers are subjected to complex environmental factors and mechanical stresses, such as high-temperature environments and short-circuit impact loads. These factors can lead to a series of performance deterioration problems, such as uneven temperature stress distribution, cumulative mechanical deformation, and insulation degradation, which in turn seriously affect the service life and operational safety of the equipment.

[0003] Currently, the industry typically employs measures such as regular inspections and preventative maintenance to ensure the reliability of converter transformers. These measures include: using infrared thermal imaging technology to monitor the temperature field distribution of the equipment; using vibration analysis methods to diagnose the mechanical condition of components; and measuring insulation resistance and leakage current to assess insulation performance. However, these methods often only obtain current static performance data and cannot fully reflect the cumulative impact of temperature and mechanical deformation on the overall performance of the equipment. Furthermore, these inspection methods usually require shutdown for inspection, making real-time monitoring of the converter transformer's operating status impossible.

[0004] In addition, some scholars have conducted specialized research on the temperature effects, short-circuit mechanical characteristics, and insulation performance of converter transformers. For example, some scholars have proposed a thermo-mechanical coupling model based on finite element analysis to calculate the stress distribution under temperature fields; others have used experimental testing methods to study the dynamic response of transformer components under short-circuit impact loads. However, most of these studies focus on a single physical field and lack a comprehensive consideration and quantitative analysis of the coupled effects of temperature, mechanical deformation, and electrical performance.

[0005] Therefore, there is an urgent need to establish a detection method that can comprehensively evaluate the interaction between temperature, mechanical deformation, and electrical performance, so as to provide a reliable basis for condition monitoring and life management of converter transformers. Summary of the Invention

[0006] In view of this, the present invention provides a method for detecting the impact of temperature and mechanical deformation accumulation on the short-circuit withstand performance of converter transformers, which can solve the problem that the existing technology lacks a detection method that can comprehensively evaluate the interaction between temperature, mechanical deformation and electrical performance.

[0007] This invention is implemented as follows:

[0008] This invention provides a method for detecting the impact of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer, comprising the following steps:

[0009] S01. Obtain the material characteristic parameters of the converter transformer, including the thermal expansion coefficient of the conductor material, the thermal expansion coefficient of the insulation material, the elastic modulus parameter of the conductor material, the elastic modulus parameter of the insulation material, the fracture toughness parameter of the conductor material, the fracture toughness parameter of the insulation material, the yield strength parameter of the conductor material, and the yield strength parameter of the insulation material.

[0010] S02. Install the converter transformer under test component on the dynamic mechanical tensile press, and heat the converter transformer under test component to a preset temperature through a temperature control system;

[0011] S03. Use a three-dimensional laser scanner to obtain the initial state parameters of the converter transformer under test component at the preset temperature. The initial state parameters include winding size parameters, pad force parameters, support bar force parameters, and insulation layer parameters.

[0012] S04. Use a temperature sensor array to measure the temperature field distribution data of the converter transformer component under test;

[0013] S05. Using finite element analysis software, calculate the initial electric field distribution parameters of the converter transformer component under test based on the temperature field distribution data and the material property parameters;

[0014] S06. Apply dynamic force load to the converter transformer component under test using the dynamic mechanical tensile press according to the preset short-circuit impact load parameters.

[0015] S07. The mechanical response data of the converter transformer under test component under the action of the dynamic force load is collected by a displacement sensor array. The mechanical response data includes the winding deformation, pad deformation, support bar deformation, and insulation layer deformation.

[0016] S08. A high-speed camera system is used to record the residual deformation data of the converter transformer under test after the dynamic force load is applied. The residual deformation data includes the residual deformation of the winding, the residual deformation of the pad, the residual deformation of the support bar, and the residual deformation of the insulation layer.

[0017] S09. Use an insulation distance measuring instrument to measure the change in insulation distance of the component under test of the converter transformer;

[0018] S10. Adjust the preset temperature and repeat steps S03 to S09 to obtain the mechanical response data, residual deformation data and insulation distance change of the converter transformer under test component at different temperatures.

[0019] S11. Establish a dynamic characteristic database of converter transformer under the cumulative effect of temperature and mechanical deformation based on the mechanical response data, the residual deformation data and the insulation distance change;

[0020] S12. Analyze the variation law of mechanical properties of conductors and winding components based on the converter transformer dynamic characteristic database;

[0021] S13. Based on the aforementioned mechanical performance variation law, establish a correlation model for the mechanical and electrical performance of the converter transformer under the cumulative effect of temperature and mechanical deformation;

[0022] S14. Based on the converter transformer dynamic characteristic database, the data is divided into a training set, a validation set, and a test set in a ratio of 8:1:1. The physical model parameters and neural network compensation terms of each equation in the association model are trained, and the trained association model is output and saved.

[0023] Based on the above technical solution, the method for detecting the influence of temperature and mechanical deformation accumulation on the short-circuit withstand performance of converter transformers in this invention can be further improved as follows:

[0024] The correlation model includes the temperature stress equation, the deformation accumulation equation, the mechanical strength equation, and the electrical performance equation.

[0025] Furthermore, the temperature stress equation is used to calculate the thermal stress distribution of the converter transformer components at different temperatures. The inputs include the temperature field distribution data, the thermal expansion coefficient of the conductor material, the thermal expansion coefficient of the insulation material, the elastic modulus parameter of the conductor material, and the elastic modulus parameter of the insulation material. The output is the component stress distribution data. The temperature stress equation includes a neural network compensation term based on a three-layer perceptron to compensate for nonlinear temperature stress effects.

[0026] Furthermore, the deformation accumulation equation is used to calculate the cumulative deformation of the component under short-circuit impact. The inputs include the stress distribution data of the component, the preset short-circuit impact load parameters, the yield strength parameters of the conductor material, and the yield strength parameters of the insulation material. The output is the cumulative deformation of the component. The deformation accumulation equation includes a neural network compensation term based on a three-layer perceptron to compensate for the nonlinear deformation accumulation effect.

[0027] Furthermore, the mechanical strength equation is used to evaluate the strength margin of the component. The inputs include the cumulative deformation of the component, the fracture toughness parameter of the conductor material, and the fracture toughness parameter of the insulation material. The output is the remaining strength value of the component. The mechanical strength equation includes a neural network compensation term based on a three-layer perceptron to compensate for the nonlinear strength evolution effect.

[0028] Furthermore, the electrical performance equation is used to calculate the impact of mechanical deformation on electrical characteristics. The inputs include the residual strength value of the component, the change in insulation distance, and the electric field distribution parameters. The output is the insulation margin of the converter transformer. The electrical performance equation includes a neural network compensation term based on a three-layer perceptron to compensate for nonlinear insulation characteristic effects.

[0029] Furthermore, the step of training the physical model parameters and neural network compensation terms of each equation in the correlation model includes:

[0030] The temperature stress equation was trained, the parameters of the physical model were fitted using the least squares method, and the parameters of the neural network compensation term were trained using the backpropagation algorithm.

[0031] The deformation cumulative equations are trained, the parameters of the physical model are fitted using the least squares method, and the parameters of the neural network compensation terms are trained using the backpropagation algorithm.

[0032] The mechanical strength equation is trained, the parameters of the physical model are fitted using the least squares method, and the parameters of the neural network compensation term are trained using the backpropagation algorithm.

[0033] The electrical performance equations are trained, the parameters of the physical model are fitted using the least squares method, and the parameters of the neural network compensation terms are trained using the backpropagation algorithm.

[0034] Furthermore, the physical model parameters refer to the collective term for material mechanical property parameters, geometric dimension parameters, and boundary condition parameters.

[0035] Furthermore, for each neural network compensation term, a lightweight three-layer perceptron structure is adopted, which is built based on the GhostNet model. An adaptive feature fusion module is added between the first Ghost module and the second Ghost module of the GhostNet model. The adaptive feature fusion module includes a channel attention submodule, a spatial attention submodule, and a discriminant equation.

[0036] The core of the GhostNet network structure is the Ghost Module, whose basic idea is to generate more feature maps from a small number of original feature maps through linear transformations: First, ordinary convolutions are used to generate a small number of primary feature maps. Then, a series of linear transformation operations (such as 1×1 convolutions, 3×3 depthwise convolutions, simple linear transformations, etc.) are applied to these feature maps to generate Ghost feature maps. These Ghost feature maps are then concatenated with the original feature maps to form the final output. The entire network starts with a standard 2D convolutional layer, followed by multiple Ghost bottlenecks. The main structure consists of a Bottleneck layer, with each Ghost bottleneck layer containing two Ghost modules and shortcut connections added at appropriate locations. The network ends with global average pooling and fully connected layers. The linear transformation introduced in the Ghost modules can greatly reduce the number of parameters and computational complexity because it does not need to learn convolutional kernels separately for each feature map. Instead, it derives new feature maps from the original feature maps through inexpensive linear operations. This design allows GhostNet to significantly reduce model size and computational overhead while maintaining good performance, making it particularly suitable for use in resource-constrained scenarios.

[0037] The neural network compensation term based on the GhostNet model offers the following advantages and effects: GhostNet's Ghost module, through linear transformation, generates more feature maps from a small number of original feature maps, effectively reducing the number of parameters and computational complexity. This is crucial for scenarios like converter transformers that require real-time monitoring and rapid response. In this scheme, the four equations—temperature stress, deformation accumulation, mechanical strength, and electrical performance—all contain nonlinear effects. Traditional neural networks may require deep network structures to fit these nonlinear relationships well, while GhostNet, through its unique feature reuse mechanism, can achieve similar fitting results with fewer parameters. Furthermore, the added adaptive feature fusion module dynamically adjusts the importance of features based on material property parameters, enabling the model to better adapt to transformer performance prediction under different material and operating conditions. Combined with the evaluation mechanism of the discriminant equation, the entire compensation system maintains low computational overhead while accurately capturing various nonlinear physical effects, thereby improving the accuracy and practicality of converter transformer short-circuit performance detection.

[0038] Furthermore, the discriminant equation is used to evaluate the matching degree between material property parameters and attention weights. The input includes the material property parameters, channel attention weights, and spatial attention weights, and the output is the effectiveness score of feature fusion.

[0039] The models and equations involved in this invention are described in detail below:

[0040] 1. Temperature field distribution calculation model:

[0041] The calculation of the temperature field distribution data is specifically represented as follows:

[0042] ;

[0043] In the formula, For spatial points In time Temperature value at any given time; The initial temperature; For the first Temperature weighting coefficient for each temperature measurement point; This is the temperature decay coefficient; For the first The spatial coordinates of each temperature measurement point; This is the time-related influence coefficient. This represents the number of temperature sensors.

[0044] Parameter acquisition method: The temperature is measured directly by a temperature sensor, and the unit is Kelvin (K). The values ​​were obtained by least squares fitting and range from 0 to 1. The values ​​were obtained through experimental calibration, and the range is 0.001~0.1. The values ​​were obtained by fitting time-series data, and the range is -0.1 to 0.1 K / s.

[0045] 2. Temperature stress equation:

[0046] The temperature stress equation is specifically expressed as follows:

[0047] ;

[0048] In the formula, It is the thermal stress tensor; It is the elastic modulus tensor; This is the thermal expansion coefficient tensor; This refers to the change in temperature. For mechanical strain tensor; This is a neural network compensation term.

[0049] Neural network compensation term The calculation formula is:

[0050] ;

[0051] In the formula, This is the weight matrix; It is the bias vector; For activation functions; This is the temperature field vector; This is a vector of thermal expansion coefficients; This is the elastic modulus vector.

[0052] 3. Cumulative Equation of Deformation:

[0053] The deformation cumulative equation is specifically expressed as follows:

[0054] ;

[0055] In the formula, For the total deformation tensor; It is the elastic deformation tensor; For plastic deformation tensor; For creep deformation tensor; This is a neural network compensation term.

[0056] The calculation formulas for each component are as follows:

[0057] ;

[0058] ,when hour;

[0059] ;

[0060] In the formula, For stress tensor; It is the elastic modulus; The plasticity coefficient; Yield strength; The work hardening index; These are material constants; Stress index; For time; For time index; To activate energy; It is the gas constant; This refers to absolute temperature.

[0061] 4. Mechanical strength equation:

[0062] The mechanical strength equation is specifically expressed as follows:

[0063] ;

[0064] ;

[0065] In the formula, Remaining strength; Initial strength; Degree of damage; This is a neural network compensation term; For the first The strain amplitude of the next cycle; For fracture strain; These are material-related constants.

[0066] 5. Electrical performance equations:

[0067] The electrical performance equation is specifically expressed as follows:

[0068] ;

[0069] In the formula, For insulation margin; This is the initial insulation margin; The attenuation coefficient; This represents the change in insulation distance; The electric field intensity vector; This is a neural network compensation term.

[0070] 6. Discriminant equation:

[0071] The discriminant equation is specifically expressed as follows:

[0072] ;

[0073] In the formula, The score is used to measure the effectiveness of feature fusion. These are the weighting coefficients; Channel attention score; Score for spatial attention; This is a material property mapping function; This is a vector of material property parameters.

[0074] Explanation of the principle:

[0075] 1. The temperature field distribution model adopts the superposition of Gaussian kernel functions, which takes into account the continuous spatial distribution characteristics and temporal evolution features of temperature;

[0076] 2. The temperature stress equation is based on thermoelastic theory, and a neural network term is introduced to compensate for nonlinear effects;

[0077] 3. The deformation accumulation equation comprehensively considers the three deformation mechanisms of elasticity, plasticity and creep, and is suitable for long-term service conditions at high temperatures.

[0078] 4. The mechanical strength equation is based on the theory of continuous damage mechanics and introduces the concept of cumulative damage;

[0079] 5. The electrical performance equations use an exponential decay model to describe the insulation performance degradation process;

[0080] 6. The discriminant equation uses a weighted fusion method to evaluate the feature extraction effect.

[0081] The derivation process of each equation is described in detail below:

[0082] 1. Derivation of the temperature field distribution model:

[0083] Step 1: Based on Fourier's law of heat conduction, establish the basic heat conduction equation:

[0084] ;

[0085] In the formula, The temperature diffusion coefficient, measured in m² / s, can be calculated from the material's thermal conductivity, density, and specific heat capacity.

[0086] Step 2: Considering the influence of the heat source term, extend the heat conduction equation:

[0087] ;

[0088] In the formula, This is the volumetric heat source density, with units of W / m³. This refers to the material density, expressed in kg / m³. Specific heat capacity, expressed in J / (kg·K).

[0089] Step 3: Solve the partial differential equation using the Green's function method to obtain the temperature field expression:

[0090] .

[0091] 2. Derivation of the temperature stress equation:

[0092] Step 1: Establish the stress-strain relationship based on thermoelasticity theory:

[0093] ;

[0094] In the formula, For total strain; For elastic strain; This is thermal strain.

[0095] Step 2: Substitute Hooke's law and the expression for thermal strain:

[0096] ;

[0097] In the formula, It is the elastic modulus; The coefficient of thermal expansion; This represents the change in temperature.

[0098] Step 3: Considering the effects of mechanical deformation, introduce a mechanical strain term:

[0099] .

[0100] Step 4: Introduce a neural network compensation term to handle nonlinear effects:

[0101] .

[0102] 3. Derivation of the cumulative deformation equation:

[0103] Step 1: Based on the principle of strain decomposition, list the expression for the total deformation:

[0104] .

[0105] Step 2: Establish the elastic deformation expression based on Hooke's Law: .

[0106] Step 3: Establish the expression for plastic deformation based on the Ramberg-Osgood model:

[0107] ,when hour.

[0108] Step 4: Establish the creep deformation expression based on the Norton creep model:

[0109] .

[0110] Step 5: Introduce neural network compensation terms:

[0111] .

[0112] 4. Derivation of the mechanical strength equation:

[0113] Step 1: Based on the theory of continuous damage mechanics, establish the damage evolution equation: ;

[0114] In the formula, This represents the number of loop iterations. For damage variables; This refers to the strain amplitude; For fracture strain; is a material constant.

[0115] Step 2: Integrate to obtain the cumulative damage expression: .

[0116] Step 3: Establish the relationship between strength and damage: ;

[0117] In the formula, Remaining strength; This represents the initial strength.

[0118] Step 4: Introduce a neural network compensation term: .

[0119] 5. Derivation of electrical performance equations:

[0120] Step 1: Based on electric field theory, establish the relationship between electric field strength and insulation distance: ;

[0121] In the formula, Voltage; This is the insulation distance.

[0122] Step 2: Consider the impact of changes in insulation distance: ;

[0123] In the formula, This represents the change in insulation distance.

[0124] Step 3: Establish an insulation margin attenuation model: ;

[0125] In the formula, For insulation margin; This is the initial insulation margin; This is the attenuation coefficient.

[0126] Step 4: Introduce a neural network compensation term: .

[0127] 6. Derivation of the discrimination equation:

[0128] Step 1: Calculate the channel attention score: ;

[0129] In the formula, This is the channel weight vector; For channel feature vectors; This represents the number of channels.

[0130] Step 2: Calculate the spatial attention score: ;

[0131] In the formula, This is the spatial weight vector; For spatial feature vectors; This represents the number of spatial locations.

[0132] Step 3: Construct the material property mapping function: ;

[0133] In the formula, This is the weight matrix; It is the bias vector; This is a vector of material property parameters.

[0134] Step 4: Weighted fusion to obtain the final discriminant equation: .

[0135] Additional notes on parameter acquisition methods: Obtained through material handbook lookup or experimental measurement; Obtained by measuring heat flux density using a heat flux density sensor; Obtained by consulting the materials manual; Obtained through tensile testing; The parameters were obtained through measurements using a thermal expansion meter; the neural network parameters were obtained through training using the backpropagation algorithm. Obtained through uniaxial tensile testing; Obtained through high-temperature creep testing; The gas constant is taken as 8.314 J / (mol·K); Obtained through tensile testing; Obtained through fatigue testing; Obtained through non-destructive testing methods; Obtained through insulation breakdown test; Obtained through dielectric strength testing; The parameters were obtained through cross-validation optimization; the neural network parameters were obtained through backpropagation (BP) training.

[0136] Specifically, the steps for obtaining the experimental parameters are as follows:

[0137] 1. Elastic modulus The acquisition steps are as follows: Sample preparation: Prepare tensile specimens conforming to GB / T228.1 standard from the conductor material and insulation material of the converter transformer; Mount the specimens on a universal testing machine and set the tensile rate to 2 mm / min; Apply a preload, eliminate the specimen clamping gap, and set the preload to 20 N; Start the test and record the stress-strain curve; Select 5 points in the linear segment of the stress-strain curve and calculate the slope using the least squares method, which is the elastic modulus. .

[0138] 2. Coefficient of thermal expansion The steps for obtaining the sample are as follows: The material sample is processed into a cylinder with a length of 25 ± 0.5 mm; the sample is placed in the test chamber of the thermal expansion apparatus; the temperature is increased from room temperature to 200℃ at a rate of 5℃ / min; the length change of the sample is recorded every 10℃; and the temperature is then determined according to the formula... Calculate the coefficient of thermal expansion; where, The initial length of the sample is in mm. This represents the change in length, expressed in mm. This represents the change in temperature, expressed in Kelvin (K).

[0139] 3. Plasticity parameters The steps for obtaining the strain are as follows: Prepare tensile specimens conforming to GB / T228.1 standard; conduct uniaxial tensile tests on a universal testing machine; record complete stress-strain curves until the specimen breaks; and then, according to the formula... Nonlinear fitting was performed on the data after the yield point; the parameters were determined using the least squares method. and The value of .

[0140] 4. Creep parameters The steps for obtaining the creep test specimen are as follows: Prepare a creep specimen conforming to GB / T2039 standard; mount the specimen on a high-temperature creep testing machine; conduct creep tests at different temperatures (e.g., 150℃, 175℃, 200℃) and different stress levels; record the creep strain variation curve over time; and apply the Norton creep formula. Multivariate nonlinear fitting was performed on data under different temperatures and stress levels to obtain parameters. The value of .

[0141] 5. Fracture strain and damage index The steps for obtaining the strain are as follows: Prepare standard fatigue specimens; conduct strain-controlled fatigue tests on a fatigue testing machine; set different strain amplitude levels, such as 0.2%, 0.4%, and 0.6%; record the fatigue life at each strain amplitude level; and obtain the fracture strain by fitting the strain according to the Manson-Coffin formula. According to the damage evolution equation obtained by fitting value.

[0142] 6. Insulation performance parameters The steps for obtaining the standard insulation breakdown test sample are as follows: Prepare a standard insulation breakdown test sample; install the sample on the insulation breakdown test apparatus; gradually increase the voltage until breakdown occurs, and record the breakdown voltage; repeat step 3 at different temperatures to obtain the relationship between temperature and breakdown voltage; and then apply the formula... Fitting to obtain parameters .

[0143] 7. Temperature distribution related parameters The acquisition steps are as follows: Install a temperature sensor array; arrange heat flux density sensors at different locations; collect temperature field distribution data and heat flux density data; calculate the temperature diffusion coefficient according to the Fourier heat conduction equation. Volumetric heat source density is obtained directly by measuring the heat flux density using a heat flux density sensor. .

[0144] 8. Weight coefficients of the discriminant equation The steps to obtain the training dataset are as follows: Prepare the training dataset, which contains feature data under different working conditions; initialize the weight coefficients to equal values; evaluate the model performance using cross-validation; optimize the weight coefficients using gradient descent; repeat steps 3 and 4 until convergence.

[0145] Precautions: All experiments must be conducted under standard environmental conditions (temperature 23±2℃, relative humidity 50±10%); each experiment must be repeated at least 3 times to ensure data reliability.

[0146] Compared with existing technologies, the beneficial effects of the method for detecting the influence of temperature and mechanical deformation accumulation on the short-circuit withstand performance of converter transformers provided by this invention are as follows: By systematically measuring the mechanical response, residual deformation, and insulation distance changes of converter transformer components under different temperatures and short-circuit impact loads, a comprehensive correlation model covering temperature stress, deformation accumulation, mechanical strength, and electrical performance is established. Compared with existing technologies, the solution of this invention has the following innovations and advantages:

[0147] 1. It achieves comprehensive detection and analysis of the coupled effects of temperature, mechanical deformation, and electrical performance, making up for the shortcomings of existing research in a single physical field. The established correlation model can fully reflect the performance degradation process of key components of converter transformers under actual service conditions, providing a reliable basis for equipment condition monitoring and life management.

[0148] 2. Advanced methods such as dynamic mechanical testing, 3D scanning, and high-speed photography were used to obtain comprehensive performance data of the converter transformer under temperature and short-circuit impact, providing reliable basic data support for the establishment of the correlation model.

[0149] 3. A neural network-based compensation term was introduced into the correlation model, which can effectively compensate for nonlinear temperature stress effects, deformation accumulation effects, strength evolution effects and insulation characteristic effects, thereby improving the prediction accuracy of the model.

[0150] 4. An adaptive feature fusion mechanism is adopted, which enables the model to dynamically adjust the importance of features according to material properties, thereby improving its ability to adapt to different working conditions and enhancing its practicality.

[0151] In summary, this invention solves the problem of the lack of a detection method in the prior art that can comprehensively evaluate the interaction between temperature, mechanical deformation, and electrical performance. Attached Figure Description

[0152] Figure 1 A flowchart of the method provided by the present invention. Detailed Implementation

[0153] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0154] like Figure 1 The diagram shows a flowchart of a method for detecting the influence of temperature and mechanical deformation accumulation on the short-circuit withstand performance of a converter transformer, provided by this invention. The specific implementation steps of this invention are described in detail below: Step S01 involves obtaining various physical performance parameters of the converter transformer materials. First, it is necessary to measure parameters such as the coefficient of thermal expansion, elastic modulus, fracture toughness, and yield strength of the conductor material. It is also necessary to measure parameters such as the coefficient of thermal expansion, elastic modulus, fracture toughness, and yield strength of the insulating material. These parameters can be obtained through experimental testing or by consulting material property handbooks. Understanding these material characteristic parameters is crucial for subsequent temperature stress analysis, deformation accumulation calculation, and mechanical strength assessment.

[0155] The specific implementation of step S02 involves mounting the converter transformer component to be tested on a dynamic mechanical testing machine and heating it to a preset temperature using a temperature control system. First, the converter transformer component is fixed to the fixture of the dynamic tensile testing machine, ensuring its position and orientation are stable. Then, the component is heated using the temperature control system. The heating temperature can be set between 60℃ and 120℃, appropriately selected according to actual service conditions. The temperature control system can employ electric heating, hot airflow heating, or infrared radiation heating, and is equipped with a temperature sensor for real-time monitoring and feedback control, ensuring the component temperature remains stable near the preset value. This method simulates the working state of a converter transformer in a high-temperature environment.

[0156] The specific implementation of step S03 involves using a 3D laser scanner to acquire the initial geometric parameters of the converter transformer components at a preset temperature. First, the component to be tested, heated to the preset temperature, is placed within the scanning area of ​​the 3D laser scanner, ensuring that the component surface is unobstructed. Then, the scanner is activated to perform a full scan, acquiring 3D point cloud data of the component surface. Professional 3D measurement software is used to process and analyze the point cloud data, extracting key geometric parameters such as winding dimensions, pad stress, support bar stress, and insulation layer thickness. These initial geometric parameters provide the foundational data for subsequent temperature stress analysis, deformation accumulation calculation, and insulation performance evaluation. The measurement accuracy of a 3D laser scanner is typically around 0.1 mm, which meets the requirements of this method.

[0157] The specific implementation of step S04 involves using a temperature sensor array to measure the temperature field distribution of the converter transformer components. First, multiple temperature sensors, such as thermocouples or resistance temperature detectors (RTDs), are arranged at key locations on the surface or inside the components, forming a temperature sensor array. These sensors should cover the main areas of the components, such as windings, pads, support bars, and insulation layers. Then, the temperature readings of each temperature sensor are recorded in real time using a data acquisition device and compiled into temperature field distribution data. Typically, the measurement accuracy of the temperature sensors can reach 0.1℃, and the number of sensors in the array depends on the complexity of the components, generally around 10-20. This temperature field distribution data provides the necessary input for subsequent temperature stress calculations.

[0158] The specific implementation of step S05 involves using finite element analysis software to calculate the initial electric field distribution of the converter transformer components based on temperature field distribution data and material parameters. First, a three-dimensional finite element model of the converter transformer components needs to be established, and the correct geometric dimensions, material properties, and boundary conditions must be set in the model. Then, the temperature field distribution data obtained in step S04 is imported into the finite element model as a thermal load condition. Using the thermo-electric coupling solver in the finite element analysis software, the initial electric field distribution of the components at a preset temperature is calculated based on parameters such as the relative permittivity and conductivity of the material. This electric field distribution information is crucial for subsequent evaluation of the impact of mechanical deformation on insulation performance. Finite element analysis software can typically provide numerical results of the electric field intensity distribution with an accuracy of approximately 10 V / mm.

[0159] The specific implementation of step S06 involves applying a preset short-circuit impact load to the converter transformer components using a dynamic mechanical testing machine. First, it is necessary to determine the typical impact load spectrum under a short-circuit accident, including parameters such as peak force and load duration. These parameters can be determined through relevant standards or engineering experience. For example, for a 500kV converter transformer, the peak force of the short-circuit impact load can be set at approximately 50kN, with a duration of 50-100ms. Then, the component to be tested, heated to the preset temperature in step S02, is mounted on the dynamic testing machine and dynamically loaded according to the preset short-circuit impact load spectrum. The testing machine should be able to provide sufficient loading capacity and response speed to simulate the actual short-circuit impact process. In this way, mechanical response data of the converter transformer components under dynamic load can be obtained.

[0160] The specific implementation of step S07 involves using a displacement sensor array to measure the mechanical response of the converter transformer components under dynamic load. First, multiple displacement sensors are deployed at key locations on the components, such as windings, pads, support bars, and insulation layers, forming a complete measurement array. These sensors should be able to accurately capture the deformation response of the components under dynamic load. Typical sensors include resistive displacement sensors, optical displacement sensors, or laser displacement sensors. Then, during the application of the dynamic load in step S06, the displacement response data at each location is recorded in real time using a data acquisition system. This data includes the deformation of the windings, pads, support bars, and insulation layers. The measurement accuracy of the sensors is typically around 0.01 mm, which meets the requirements of this method. This mechanical response data provides the basis for subsequent deformation accumulation calculations and mechanical strength assessments.

[0161] The specific implementation of step S08 involves using a high-speed camera system to record the residual deformation of the converter transformer components after the dynamic load has ended. First, the component to be tested, loaded in step S06, is placed within the imaging range of the high-speed camera system, ensuring the component surface is clearly visible. Then, the high-speed camera system is activated, continuously recording the deformation process of the component after the dynamic load has ended at a rate of at least 1000 frames per second. By analyzing these high-speed video images, residual deformation data for key components such as windings, pads, supports, and insulation layers can be extracted. The measurement accuracy of the high-speed camera system is typically up to 0.05 mm, which meets the measurement requirements of this method for residual deformation. This residual deformation data is of great significance for evaluating the mechanical strength margin and insulation performance of the components.

[0162] The specific implementation of step S09 involves using an insulation distance measuring instrument to measure the changes in insulation distance among the converter transformer components. First, insulation distance measuring instruments, such as contact or non-contact capacitive or optical distance sensors, are installed at key insulation gap locations on the components. These sensors should be able to accurately measure changes in insulation distance. Then, during and after the application of the dynamic load in step S06, the distance changes of each insulation gap are recorded in real time using a data acquisition system. The measurement accuracy can reach 0.01 mm. This insulation distance change data is a crucial basis for evaluating the impact of mechanical deformation on insulation performance.

[0163] The specific implementation of step S10 involves adjusting the preset temperature and repeating steps S03 to S09 to obtain mechanical response data, residual deformation data, and insulation distance changes of the converter transformer components at different temperatures. First, the range and step size of the temperature adjustment need to be determined. Four typical temperature points—60℃, 80℃, 100℃, and 120℃—can be selected for testing. For each temperature point, the operation process of steps S03 to S09 is repeated, including 3D scanning to obtain initial geometric parameters, temperature field distribution measurement, electric field distribution calculation, dynamic load application, mechanical response measurement, residual deformation recording, and insulation distance measurement. This method allows for the acquisition of comprehensive performance data of the converter transformer components under different temperature conditions, providing a foundation for subsequently establishing a correlation model between temperature, mechanical deformation, and electrical performance.

[0164] The specific implementation of step S11 involves establishing a dynamic characteristic database of the converter transformer based on the aforementioned mechanical response data, residual deformation data, and insulation distance changes. First, the data obtained from steps S03 to S09 are organized and archived to form a systematic database. This database should include winding size parameters, pad stress parameters, support bar stress parameters, insulation layer parameters, as well as mechanical response data, residual deformation data, and insulation distance changes under different temperature conditions. Corresponding material property parameters such as thermal expansion coefficient, elastic modulus, fracture toughness, and yield strength should also be entered into the database. This establishes a comprehensive performance database covering multiple aspects, including temperature, mechanical deformation, and electrical performance. This database provides fundamental data support for subsequent correlation model establishment and parameter training.

[0165] The specific implementation of step S12 is based on analyzing the mechanical performance variation patterns of conductors and winding components using a converter transformer dynamics characteristic database. First, the mechanical response data and residual deformation data under different temperature and load conditions in the database are analyzed and compared holistically. Typical performance variation patterns of conductors and winding components under temperature and mechanical deformation can be extracted using statistical analysis, curve fitting, and other methods. For example, the relationship between the degree of plastic deformation of the conductor material and temperature and stress, and the evolution characteristics of winding deformation with temperature and short-circuit impact force can be analyzed. Simultaneously, the influence of different material parameters such as thermal expansion coefficient and yield strength on component performance can be studied. This analysis allows for a deeper understanding of the influence mechanism of temperature and mechanical deformation on the mechanical performance of key components of the converter transformer.

[0166] The specific implementation of step S13 involves establishing a correlation model between temperature, mechanical deformation, and electrical performance based on the aforementioned mechanical property change patterns. This correlation model should include the following four sub-models:

[0167] 1. Temperature Stress Equation: Based on temperature field distribution data, material thermal expansion coefficient, elastic modulus, and other parameters, the internal stress distribution of the component at different temperatures is calculated using thermoelastic theory. A neural network term is also introduced to compensate for nonlinear temperature stress effects.

[0168] 2. Deformation Cumulative Equation: Taking into account mechanisms such as elastic deformation, plastic deformation, and creep deformation, the cumulative deformation of the component under short-circuit impact load is calculated based on parameters such as stress distribution and material yield strength. A neural network term is also introduced to compensate for the nonlinear deformation cumulative effect.

[0169] 3. Mechanical Strength Equation: Based on the theory of continuous damage mechanics, and combined with parameters such as the fracture toughness of the conductor and insulation materials, the residual mechanical strength of the component under cumulative deformation is evaluated. A neural network term is also introduced to compensate for the nonlinear strength evolution effect.

[0170] 4. Electrical performance equations: Based on parameters such as the remaining mechanical strength of the component, the change in insulation distance, and the electric field distribution, an exponential decay model is used to calculate the degree of insulation performance degradation. A neural network term is also introduced to compensate for nonlinear insulation characteristic effects.

[0171] By establishing such a comprehensive correlation model, the effects of temperature and mechanical deformation on the mechanical strength and insulation performance of key components of converter transformers can be predicted more accurately, providing a theoretical basis for equipment condition monitoring and life assessment.

[0172] The specific implementation of step S14 involves training and optimizing the established correlation model. First, the dynamic feature database established in step S11 is divided into a training set, a validation set, and a test set in an 8:1:1 ratio. Then, for the above four sub-models, the physical model parameters and neural network compensation terms are trained using the following methods:

[0173] 1. Temperature stress equation: The least squares method is used to fit the physical model parameters such as the coefficient of thermal expansion and the elastic modulus; the backpropagation algorithm is used to train the neural network compensation term parameters.

[0174] 2. Deformation cumulative equation: The least squares method is used to fit the physical model parameters such as yield strength and creep coefficient; the backpropagation algorithm is used to train the neural network compensation term parameters.

[0175] 3. Mechanical strength equation: The least squares method is used to fit the physical model parameters, such as fracture toughness; the backpropagation algorithm is used to train the neural network compensation term parameters.

[0176] 4. Electrical performance equations: The least squares method is used to fit the physical model parameters, such as the insulation distance attenuation coefficient; the backpropagation algorithm is used to train the neural network compensation term parameters.

[0177] When training the neural network compensation term, a lightweight three-layer perceptron architecture is employed, with an adaptive feature fusion module added within it. This fusion module includes a channel attention submodule, a spatial attention submodule, and a discriminant equation to evaluate the matching degree between material property parameters and attention weights, thereby dynamically adjusting the importance of features. This design can improve the model's ability to capture nonlinear physical effects while maintaining low computational overhead.

[0178] Specifically, the principle of this invention is to establish a correlation model that comprehensively considers the coupling relationship between temperature effects, mechanical deformation effects, and electrical performance. This model includes four main sub-models: a temperature stress model, a deformation accumulation model, a mechanical strength model, and an electrical performance model.

[0179] First, by measuring temperature field distribution data and material property parameters, the internal stress distribution of the component at different temperatures is calculated using thermoelastic theory. To capture the nonlinear temperature stress effect, a neural network-based compensation term is introduced into the physical model.

[0180] Secondly, a mathematical model describing the cumulative deformation of a component under dynamic force was established by combining short-circuit impact load parameters and material strength characteristics. This model comprehensively considers mechanisms such as elastic deformation, plastic deformation, and creep deformation, and also introduces a neural network term to compensate for nonlinear deformation effects.

[0181] Furthermore, based on the theory of continuous damage mechanics, the residual mechanical strength of the component under cumulative deformation was evaluated. A neural network compensation term was used to fit the nonlinear strength evolution process.

[0182] Finally, based on parameters such as the remaining mechanical strength of the component, the change in insulation distance, and the electric field distribution, an exponential decay model is used to calculate the degree of insulation performance degradation. A neural network term is also introduced to capture nonlinear insulation characteristic effects.

[0183] Throughout the establishment of the correlation model, comprehensive performance data obtained through advanced methods such as dynamic mechanical experiments, 3D scanning, and high-speed photography were fully utilized. Simultaneously, an adaptive feature fusion mechanism was introduced into the design of the neural network compensation term, enabling the model to dynamically adjust the importance of features based on material properties, thereby improving its adaptability and generalization ability.

[0184] The following is an example of a specific application scenario of the present invention: A large power grid company is operating a 500kV converter transformer substation. The transformer was put into operation in 2020 and has been in service for 5 years. In order to ensure the reliable operation of the equipment, the power grid company plans to use the temperature-mechanical deformation-electrical performance integrated testing method proposed in this invention to conduct a comprehensive diagnosis of the operating status of the converter transformer.

[0185] First, the engineering team measured and determined the key material properties of the converter transformer. The coefficient of thermal expansion of the conductor material was... elastic modulus fracture toughness Yield strength The coefficient of thermal expansion of insulating materials elastic modulus fracture toughness Yield strength These parameters will serve as the basic input data for subsequent analysis and calculations.

[0186] After mounting the converter transformer component to be tested on the dynamic mechanical testing machine, the engineering team heated it to 60°C using a temperature control system. The initial geometric parameters of the component at 60°C were obtained using a 3D laser scanner: winding length... Winding diameter Winding thickness ; Area of ​​the pad bearing force The pad block is under stress ;Strut bearing area , strut stress Insulation layer thickness .

[0187] Subsequently, the engineering team used a temperature sensor array to measure the temperature field distribution of the converter transformer component at 60°C. After data fitting, the temperature field distribution can be expressed as:

[0188] ;

[0189] in, Between 0.1 and 0.9 Between 0.01 and 0.05, The coordinates of 15 temperature sensors are shown. This temperature field distribution data was imported into finite element analysis software, and the initial electric field distribution of the component at 60℃ was calculated. The maximum value is 8kV / mm.

[0190] Based on the preset short-circuit impact load parameters, the engineering team applied peak force to the component under test using a dynamic mechanical testing machine. Duration Dynamic force load. Using a displacement sensor array, they acquired real-time mechanical response data of the component under dynamic force load: winding deformation: , , ; Deformation of the pad block: , Deformation of the support bar: , Insulation layer deformation: ;

[0191] Subsequently, the engineering team used a high-speed camera system to record the residual deformation of the component after the dynamic load was applied: Residual deformation of the winding: , , Residual deformation of the pad block: , Residual deformation of the strut: , Residual deformation of the insulation layer: ;

[0192] The change in insulation distance after a short-circuit impact was also measured using an insulation distance measuring instrument. .

[0193] Based on this comprehensive performance test data, the engineering team established a dynamic characteristic database for the converter transformer, covering indicators such as initial geometric parameters, mechanical response data, residual deformation data, and insulation distance changes. Material property parameters were also entered into this database.

[0194] Next, the engineering team began to build a correlation model between temperature, mechanical deformation, and electrical performance.

[0195] First, based on thermoelastic theory, they established the temperature stress equation:

[0196] ;

[0197] In the formula, For thermal stress tensor, For the elastic modulus tensor, For the thermal expansion coefficient tensor, The change in temperature This is the mechanical strain tensor. To capture the nonlinear temperature stress effect, This neural network compensation term was introduced into the model. After least squares fitting and backpropagation training, the physical model parameters and neural network compensation term parameters of the temperature stress equation were determined.

[0198] Secondly, taking into account elastic deformation, plastic deformation, and creep deformation, the engineering team established a deformation accumulation equation:

[0199] ;

[0200] ;

[0201] ;

[0202] ;

[0203] Also introduced A neural network term is used to compensate for the cumulative effect of nonlinear deformation. Through fitting and training, the various physical parameters and neural network parameters have been determined.

[0204] Furthermore, based on the theory of continuous damage mechanics, the engineering team established the mechanical strength equation:

[0205] ;

[0206] ;

[0207] in, For residual strength, For initial strength, This refers to the degree of damage. Similarly, Neural network compensation terms are used to fit the nonlinear intensity evolution process.

[0208] Finally, the electrical performance equations were established using an exponential decay model:

[0209] ;

[0210] here, For insulation margin, For changes in insulation distance, denoted as electric field strength. The neural network compensation term is used to capture the nonlinear insulation degradation effect.

[0211] In building the four sub-models mentioned above, the engineering team employed an adaptive feature fusion mechanism, which includes channel attention, spatial attention, and a discriminant equation for evaluating feature importance. This allows the model to dynamically adjust feature weights based on changes in material property parameters, thereby improving its adaptability and generalization ability.

[0212] Finally, the engineering team divided the entire converter transformer dynamic characteristic database into training, validation, and test sets in an 8:1:1 ratio, and trained and optimized the four sub-models mentioned above. During training, the physical model parameters were fitted using the least squares method, while the neural network compensation terms were optimized using the backpropagation algorithm. After repeated iterations, a relatively accurate temperature-mechanical deformation-electrical performance correlation model was finally obtained.

[0213] Using this correlation model, the engineering team conducted a comprehensive diagnosis of the operating status of the 500kV converter transformer:

[0214] 1. Temperature stress analysis: Maximum thermal stress inside the component at 60℃. The stress is mainly concentrated in the windings and support structure. Although this stress level is lower than the yield strength of the material, it exceeds the tensile strength of the insulation material, which can lead to early damage such as microcracks in the insulation layer.

[0215] 2. Deformation Cumulative Analysis: Under short-circuit impact load, the component undergoes significant cumulative deformation, such as radial deformation of the winding. axial deformation These deformations have exceeded the design limits and pose a threat to the winding insulation performance and the overall geometric stability of the machine.

[0216] 3. Mechanical strength assessment: The remaining mechanical strength of the component is estimated based on the damage accumulation model. The strength has dropped to 550 MPa, only 68.8% of its initial strength. This severe strength degradation greatly increases the risk of equipment failure.

[0217] 4. Insulation Performance Prediction: Based on calculations using the electrical performance model, the insulation margin of the converter transformer under the aforementioned temperature and mechanical deformation conditions is predicted. At only 1.2, it is already below the safe threshold of 1.5. A reduction in the insulation distance and an increase in the local electric field strength of the insulation layer will accelerate the aging and breakdown of the insulation material.

[0218] Based on the above analysis, the engineering team believes that the 500kV converter transformer is in a critical operating state and urgently needs to be repaired or replaced.

[0219] It should be noted that the variables involved in the description of this invention are shown in Table 1 below.

[0220] Table 1. Variable Explanation Table

[0221]

[0222] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer, characterized in that, Includes the following steps: S01. Obtain the material characteristic parameters of the converter transformer, including the thermal expansion coefficient of the conductor material, the thermal expansion coefficient of the insulation material, the elastic modulus parameter of the conductor material, the elastic modulus parameter of the insulation material, the fracture toughness parameter of the conductor material, the fracture toughness parameter of the insulation material, the yield strength parameter of the conductor material, and the yield strength parameter of the insulation material. S02. Install the converter transformer under test component on the dynamic mechanical tensile press, and heat the converter transformer under test component to a preset temperature through a temperature control system; S03. Use a three-dimensional laser scanner to obtain the initial state parameters of the converter transformer under test component at the preset temperature. The initial state parameters include winding size parameters, pad force parameters, support bar force parameters, and insulation layer parameters. S04. Use a temperature sensor array to measure the temperature field distribution data of the converter transformer component under test; S05. Using finite element analysis software, calculate the initial electric field distribution parameters of the converter transformer component under test based on the temperature field distribution data and the material property parameters; S06. Apply dynamic force load to the converter transformer component under test using the dynamic mechanical tensile press according to the preset short-circuit impact load parameters. S07. The mechanical response data of the converter transformer under test component under the action of the dynamic force load is collected by a displacement sensor array. The mechanical response data includes the winding deformation, pad deformation, support bar deformation, and insulation layer deformation. S08. A high-speed camera system is used to record the residual deformation data of the converter transformer under test after the dynamic force load is applied. The residual deformation data includes the residual deformation of the winding, the residual deformation of the pad, the residual deformation of the support bar, and the residual deformation of the insulation layer. S09. Use an insulation distance measuring instrument to measure the change in insulation distance of the component under test of the converter transformer; S10. Adjust the preset temperature and repeat steps S03 to S09 to obtain the mechanical response data, residual deformation data and insulation distance change of the converter transformer under test component at different temperatures. S11. Establish a dynamic characteristic database of converter transformer under the cumulative effect of temperature and mechanical deformation based on the mechanical response data, the residual deformation data and the insulation distance change; S12. Analyze the variation law of mechanical properties of conductors and winding components based on the converter transformer dynamic characteristic database; S13. Based on the aforementioned mechanical performance variation law, establish a correlation model for the mechanical and electrical performance of the converter transformer under the cumulative effect of temperature and mechanical deformation; S14. Based on the converter transformer dynamic characteristic database, the data is divided into a training set, a validation set, and a test set in a ratio of 8:1:

1. The physical model parameters and neural network compensation terms of each equation in the association model are trained, and the trained association model is output and saved.

2. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 1, characterized in that, The correlation model includes the temperature stress equation, the deformation accumulation equation, the mechanical strength equation, and the electrical performance equation.

3. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 2, characterized in that, The temperature stress equation is used to calculate the thermal stress distribution of converter transformer components at different temperatures. The inputs include the temperature field distribution data, the thermal expansion coefficient of the conductor material, the thermal expansion coefficient of the insulation material, the elastic modulus parameter of the conductor material, and the elastic modulus parameter of the insulation material. The output is the stress distribution data of the components. The temperature stress equation includes a neural network compensation term based on a three-layer perceptron to compensate for nonlinear temperature stress effects.

4. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 3, characterized in that, The deformation accumulation equation is used to calculate the cumulative deformation of the component under short-circuit impact. The inputs include the stress distribution data of the component, the preset short-circuit impact load parameters, the yield strength parameters of the conductor material, and the yield strength parameters of the insulation material. The output is the cumulative deformation of the component. The deformation accumulation equation includes a neural network compensation term based on a three-layer perceptron to compensate for the nonlinear deformation accumulation effect.

5. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 4, characterized in that, The mechanical strength equation is used to evaluate the strength margin of the component. The inputs include the cumulative deformation of the component, the fracture toughness parameter of the conductor material, and the fracture toughness parameter of the insulation material. The output is the remaining strength value of the component. The mechanical strength equation includes a neural network compensation term based on a three-layer perceptron to compensate for the nonlinear strength evolution effect.

6. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 5, characterized in that, The electrical performance equation is used to calculate the impact of mechanical deformation on electrical characteristics. The inputs include the residual strength value of the component, the change in insulation distance, and the electric field distribution parameters. The output is the insulation margin of the converter transformer. The electrical performance equation includes a neural network compensation term based on a three-layer perceptron to compensate for nonlinear insulation characteristic effects.

7. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 6, characterized in that, The step of training the physical model parameters and neural network compensation terms of each equation in the correlation model includes: The temperature stress equation was trained, the parameters of the physical model were fitted using the least squares method, and the parameters of the neural network compensation term were trained using the backpropagation algorithm. The deformation cumulative equations are trained, the parameters of the physical model are fitted using the least squares method, and the parameters of the neural network compensation terms are trained using the backpropagation algorithm. The mechanical strength equation is trained, the parameters of the physical model are fitted using the least squares method, and the parameters of the neural network compensation term are trained using the backpropagation algorithm. The electrical performance equations are trained, the parameters of the physical model are fitted using the least squares method, and the parameters of the neural network compensation terms are trained using the backpropagation algorithm.

8. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 6, characterized in that, The physical model parameters refer to the collective term for material mechanical property parameters, geometric dimension parameters, and boundary condition parameters.

9. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 8, characterized in that, For each neural network compensation term, a lightweight three-layer perceptron structure is adopted, which is built based on the GhostNet model. An adaptive feature fusion module is added between the first Ghost module and the second Ghost module of the GhostNet model. The adaptive feature fusion module includes a channel attention submodule, a spatial attention submodule, and a discriminant equation.

10. The method for detecting the influence of accumulated temperature and mechanical deformation on the short-circuit withstand performance of a converter transformer according to claim 9, characterized in that, The discriminant equation is used to evaluate the matching degree between material property parameters and attention weights. The input includes the material property parameters, channel attention weights, and spatial attention weights, and the output is the effectiveness score of feature fusion.

Images