A method and device for accelerating prediction of corona resistance life of electromagnetic wire insulation layer

Abstract

The present application belongs to the field of data processing, and in particular, relates to a method and device for accelerating the prediction of the corona resistance life of an electromagnetic wire insulation layer, the method comprising: for a test sample in the electromagnetic wire insulation layer, collecting the conductivity current density and the applied electric field strength of the test sample at different temperatures; determining the electrical aging threshold corresponding to the test sample based on the conductivity current density and the applied electric field strength at different temperatures; performing an accelerated corona resistance test based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments; constructing an accelerated prediction model based on the short-time corona resistance life test value; inputting the operating environment parameters of the target sample into the accelerated prediction model to obtain the corresponding target corona resistance life. An accelerated prediction model is constructed through the short-time corona resistance life test value, and the accelerated prediction model is used to predict the corresponding corona resistance life in real time, thereby improving the efficiency of corona resistance life prediction and the accuracy of corona resistance life prediction values.

Description

Technical Field

[0001] The present application belongs to the field of data processing, and in particular relates to a method for accelerating the prediction of the corona resistance life of an electromagnetic wire insulation layer and a related device. Background Art

[0002] Electromagnet wire (also called winding wire or magnet wire) is usually made of conductive metal (such as copper or aluminum), and its surface is covered with an insulating layer to isolate the current and ensure that the conductor can work safely and efficiently in the equipment. Electromagnet wire insulation is a layer of insulating material wrapped around the outside of the magnet wire conductor, which is used in coils and windings in electrical equipment to prevent current leakage, short circuits or other electrical failures.

[0003] The corona resistance of the insulation layer of the magnet wire is a key factor in determining the service life of the magnet wire. Therefore, it is extremely important to predict the corona resistance life of the magnet wire insulation layer. In the corona resistance life assessment of the magnet wire insulation layer, since the insulation material needs to maintain its electrical properties for a long time (usually up to decades), the traditional long-term testing method faces huge challenges. Specifically, these challenges include long time, high cost, and the test conditions may not fully match the actual use environment.

[0004] In summary, it is urgent to propose a new technical solution to solve at least one of the above technical problems so as to improve the safety of electromagnetic wire service. Summary of the invention

[0005] The present application provides a method and related device for accelerating prediction of the corona resistance life of the insulation layer of an electromagnetic wire, so as to improve the efficiency of accelerating prediction of the corona resistance life of the insulation layer of the electromagnetic wire and improve the accuracy of the predicted value of the corona resistance life.

[0006] In a first aspect, the present application provides a method for accelerating the prediction of the corona resistance life of an electromagnetic wire insulation layer, which is applied to the scenario of accelerating the prediction of the corona resistance life of an electromagnetic wire insulation layer; the method for accelerating the prediction of the corona resistance life of an electromagnetic wire insulation layer comprises:

[0007] For a test sample in an electromagnetic wire insulation layer, collecting the electrical conduction current density and applied electric field strength of the test sample at different temperatures; the test sample corresponds to at least one type of thin film material;

[0008] Determining the electrical aging threshold corresponding to the test sample based on the electrical conduction current density and the applied electric field strength at different temperatures;

[0009] Performing an accelerated corona resistance test based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments; wherein the multiple test environments respectively use different temperatures and different applied electric field strengths; the applied electric field strengths used in the multiple test environments are all less than the target applied electric field strength corresponding to the electrical aging threshold;

[0010] Constructing an accelerated prediction model based on the short-time corona resistance life test value;

[0011] The operating environment parameters of the target sample are input into the accelerated prediction model to obtain the target corona resistance life corresponding to the target sample; wherein the target sample corresponds to at least one thin film material to be tested; and the operating environment parameters include at least the real-time temperature and the real-time applied electric field strength in the operating environment of the target sample.

[0012] In a second aspect, the present application provides an accelerated prediction system for the corona resistance life of an electromagnetic wire insulation layer, the accelerated prediction system for the corona resistance life of an electromagnetic wire insulation layer comprising:

[0013] A collection unit is configured to collect the electrical conduction current density and applied electric field strength of a test sample in the electromagnetic wire insulation layer at different temperatures; the test sample corresponds to at least one thin film material to be tested;

[0014] a determination unit configured to determine an electrical aging threshold corresponding to the test sample based on the electrical conduction current density and the applied electric field strength at different temperatures;

[0015] A testing unit is configured to perform an accelerated corona resistance test based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments; wherein the multiple test environments respectively use different temperatures and different applied electric field strengths; and the applied electric field strengths used in the multiple test environments are all less than the target applied electric field strength corresponding to the electrical aging threshold;

[0016] A construction unit configured to construct an accelerated prediction model based on the short-time corona resistance life test value;

[0017] The prediction unit is configured to input the operating environment parameters of the target sample into the accelerated prediction model to obtain the target corona resistance life corresponding to the target sample; wherein the target sample corresponds to at least one thin film material to be tested; and the operating environment parameters at least include the real-time temperature and the real-time applied electric field strength in the operating environment of the target sample.

[0018] In a third aspect, the present application provides an accelerated prediction device for the corona resistance life of an electromagnetic wire insulation layer, the accelerated prediction device for the corona resistance life of an electromagnetic wire insulation layer comprising:

[0019] at least one processor, memory, and input-output unit;

[0020] The memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the method for accelerating prediction of the corona resistance life of the electromagnetic wire insulation layer of the first aspect.

[0021] In a fourth aspect, the present application provides a computer-readable storage medium, which includes instructions. When the instructions are executed on a computer, the computer executes the method for accelerating the prediction of the corona resistance life of an electromagnetic wire insulation layer according to the first aspect.

[0022] The technical solution provided by the present application, in the accelerated prediction scheme of the corona resistance life of the electromagnetic wire insulation layer, for the test sample in the electromagnetic wire insulation layer, the conductivity current density and the applied electric field strength of the test sample at different temperatures are collected; the electrical aging threshold corresponding to the test sample is determined based on the conductivity current density and the applied electric field strength at different temperatures; based on the electrical aging threshold, an accelerated corona resistance test is performed to obtain the short-time corona resistance life test value of the test sample under multiple test environments; an accelerated prediction model is constructed based on the short-time corona resistance life test value; the operating environment parameters of the target sample are input into the accelerated prediction model to obtain the target corona resistance life corresponding to the target sample. This technical solution constructs an accelerated prediction model through the short-time corona resistance life test value, and uses the accelerated prediction model to predict the corresponding corona resistance life in real time, thereby improving the efficiency of the accelerated prediction of the corona resistance life of the electromagnetic wire insulation layer and the accuracy of the corona resistance life prediction value. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0024] Figure 1 It is a schematic flow chart of a method for accelerating prediction of corona resistance life of an electromagnetic wire insulation layer according to an embodiment of the present application;

[0025] Figure 2 It is a schematic diagram of the principle of an accelerated prediction method for the corona resistance life of an electromagnetic wire insulation layer according to an embodiment of the present application;

[0026] Figure 3 It is a structural schematic diagram of an accelerated prediction system for corona resistance life of an electromagnetic wire insulation layer according to an embodiment of the present application;

[0027] Figure 4 It is a structural schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0029] The embodiments of the present application provide a method for accelerating the prediction of the corona resistance life of the insulation layer of an electromagnetic wire and a related device.

[0030] Specifically, the accelerated prediction scheme for the corona resistance life of the electromagnetic wire insulation layer can be applied to the accelerated prediction system for the corona resistance life of the electromagnetic wire insulation layer. In the accelerated prediction method for the corona resistance life of the electromagnetic wire insulation layer, first, for the test sample in the electromagnetic wire insulation layer, the conductivity current density and the applied electric field strength of the test sample at different temperatures are collected. The test sample corresponds to at least one type of film material. Then, the electrical aging threshold corresponding to the test sample is determined based on the conductivity current density and the applied electric field strength at different temperatures. Next, an accelerated corona resistance test is performed based on the electrical aging threshold to obtain the short-term corona resistance life test value of the test sample under multiple test environments. Among them, the multiple test environments respectively use different temperatures and different applied electric field strengths; the applied electric field strengths used in the multiple test environments are all less than the target applied electric field strength corresponding to the electrical aging threshold. Then, an accelerated prediction model is constructed based on the short-term corona resistance life test value. Finally, the operating environment parameters of the target sample are input into the accelerated prediction model to obtain the corresponding target corona resistance life. Wherein, the target sample corresponds to at least one thin film material to be tested; the operating environment parameters at least include the real-time temperature and the real-time applied electric field strength in the operating environment where the target sample is located.

[0031] The above technical solution can also be executed by an electronic device for data processing, which can be a server, a server cluster, or a cloud server. The electronic device can also be a terminal device such as a mobile phone, a computer, or a dedicated device (such as a dedicated terminal device with an accelerated prediction system for the corona resistance life of the electromagnetic wire insulation layer). These electronic devices can also be installed with a service program for executing the accelerated prediction solution for the corona resistance life of the electromagnetic wire insulation layer.

[0032] In the embodiment of the present application, the accelerated prediction system for the corona resistance life of the electromagnetic wire insulation layer is mainly responsible for the storage of various relevant data such as input data, calculation results, observation data, visualization data, etc. collected in real time in the test environment of the electromagnetic wire insulation layer. These data may come from different applications and need to be uniformly managed and stored for subsequent analysis and processing. Exemplarily, the test environment monitoring platform for the electromagnetic wire insulation layer can be a virtual platform built on the cloud, or it can be a hardware platform built through hardware circuits or electronic components and deployed in the accelerated prediction equipment for the corona resistance life of the electromagnetic wire insulation layer (such as electromagnetic wire insulation layer production equipment, electromagnetic wire insulation layer testing equipment, etc.).

[0033] Figure 1 A schematic diagram of an accelerated prediction method for the corona resistance life of an electromagnetic wire insulation layer provided in an embodiment of the present application, such as Figure 1 , the method comprises the following steps:

[0034] 101. For a test sample in an insulation layer of an electromagnetic wire, collect the electrical conduction current density and applied electric field strength of the test sample at different temperatures;

[0035] 102, determining an electrical aging threshold corresponding to the test sample based on the electrical current density and the applied electric field strength at different temperatures;

[0036] 103, performing an accelerated corona resistance test based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments;

[0037] 104, constructing an accelerated prediction model based on the short-time corona resistance life test value;

[0038] 105 , inputting the operating environment parameters of the target sample into the accelerated prediction model to obtain the target corona resistance life corresponding to the target sample.

[0039] In the embodiment of the present application, the test sample corresponds to at least one type of film material. The film material used in the insulation layer of the electromagnetic wire is mainly to provide insulation protection and improve the heat resistance, corona resistance, mechanical strength and chemical stability of the electromagnetic wire. The types of film materials involved in the embodiments of the present application include but are not limited to: polyimide (PI), polyester (PET), polyester-imide (Polyester-imide), polytetrafluoroethylene (PTFE), polyetherimide (PEI), polyethylene naphthalate (PEN), polyamide-imide (PAI), polyvinylidene fluoride (PVDF) and polyethylene (PE) and polypropylene (PP) films.

[0040] Among them, polyimide film has excellent high temperature resistance, corona resistance and radiation resistance, and is often used for electromagnetic wire insulation in high temperature environments, such as aerospace and high-performance motors. Polyester film has good insulation performance and mechanical strength, and is widely used in household appliances, ordinary motors and power transformers. Polyesterimide film has excellent heat resistance and corona resistance, and is suitable for electromagnetic wire insulation in medium and high temperature environments, such as high-efficiency motors and transformer coils. Polytetrafluoroethylene film has excellent heat resistance and chemical inertness, and is often used for electromagnetic wire insulation in high temperature corrosive environments. Polyetherimide film has good mechanical strength, heat resistance and corona resistance, and is widely used in high-demand electronic and power equipment. Polyethylene naphthalate film has high heat resistance and electrical properties, and is often used as the insulation layer of motors and transformers. Polyamide-imide film maintains excellent insulation performance in high temperature environments and is used in high-performance motors and transformers. Polyvinylidene fluoride film has heat resistance and electrical insulation properties, and is often used for motor insulation that requires resistance to chemical corrosion and high temperature environments. Polyethylene and polypropylene films are used in the insulation of low-voltage electromagnetic wires, such as household appliances and low-power motors, due to their good insulation properties and mechanical strength.

[0041] In 101, for the test sample in the electromagnetic wire insulation layer, the conduction current density and the applied electric field strength of the test sample at different temperatures are collected. Specifically, for the test sample in the electromagnetic wire insulation layer, the process of collecting the conduction current density and the applied electric field strength at different temperatures is to evaluate the electrical properties of the insulating material under different working conditions, especially its electrical aging resistance and corona resistance.

[0042] Specifically, the conductivity current density and the applied electric field strength are important indicators that reflect the electrical properties of materials. The conductivity current density refers to the conductive current passing through a unit area, measured in amperes per square meter (A / m²), which reflects the conductivity of the material under the action of an electric field. In the electromagnetic wire insulation layer, the ideal film material should have good insulation properties, that is, under the action of a high electric field, its conductivity current density should be as low as possible. Excessive conductivity current density will cause material leakage, affect the insulation effect, and accelerate material aging. Therefore, by measuring the conductivity current density at different temperatures, it is possible to evaluate whether the material will experience increased conductivity and leakage in the actual working environment, and then judge its insulation performance and stability. For example, in a certain test environment, if the applied electric field strength is 10 kV / mm and the temperature is 150°C, the conductivity current density measured at this time is 10⁻ 7 A / m². This indicates that under higher temperatures and electric fields, the conductivity of the material remains at a low level and the insulation performance is good. If the electrical current density suddenly increases, it may indicate the aging of the material or the risk of electrical breakdown.

[0043] The applied electric field strength refers to the electric field applied by the outside world on the insulating material, usually measured in kilovolts per millimeter (kV / mm). It is an important factor in testing the electrical strength of the material. The applied electric field strength is closely related to the conductive current density. The greater the electric field strength, the greater the conductive current density generated in the material. If the electric field strength exceeds the tolerance threshold of the material, it may cause electrical aging, insulation layer damage, and even corona discharge. By controlling the combination of different temperatures and applied electric field strengths, the test can simulate the performance of the material under actual working conditions, thereby evaluating its durability and electrical safety. For example, at a temperature of 100°C, the applied electric field strength is 8 kV / mm, and the measured conductive current density is 10⁻ 8 A / m², and when the temperature is increased to 200℃ and the applied electric field strength is increased to 14 kV / mm, the conductive current density increases to 10⁻ 6 A / m². This trend indicates that as the temperature and electric field strength increase, the conductivity of the material gradually increases, which may lead to insulation failure.

[0044] In actual testing, the insulation layer material of the electromagnetic wire will be placed under different temperature conditions and different electric field strengths will be applied. For example, the test is first carried out in a low temperature environment (such as 80°C) and a low electric field (such as 5 kV / mm), and then gradually increased to a medium temperature (such as 150°C) and a medium electric field strength (such as 10 kV / mm), and finally to a high temperature (such as 200°C) and a high electric field strength (such as 15 kV / mm). Under each condition, the change in the electrical conductivity and current density of the material is recorded. Through these data, the change trend of the electrical insulation performance of the material with temperature and electric field can be analyzed, and its long-term reliability and aging behavior in actual work can be predicted.

[0045] In general, by collecting the conductive current density and applied electric field strength at different temperatures, the electrical aging resistance of thin film materials can be evaluated in detail, and a basis can be provided for the selection and improvement of materials.

[0046] As an optional embodiment, in 102, the electrical aging threshold corresponding to the test sample is determined based on the electrical current density and the applied electric field strength at different temperatures, such as Figure 2 As shown, it can be implemented as:

[0047] 201, inputting the conductivity current density and applied electric field strength of the test sample at different temperatures into a relationship prediction model, and fitting to obtain a relationship curve between the conductivity current density and applied electric field strength corresponding to the test sample;

[0048] 202 , identifying a jump position in the relationship curve, and using the applied electric field intensity corresponding to the jump position as the electrical aging threshold.

[0049] In an embodiment of the present application, the relationship prediction model matches the type of film material to which the test sample belongs. In this embodiment, the relationship prediction model is used to determine the electrical aging threshold of the test sample based on the conductivity current density and the applied electric field strength at different temperatures. The core is to obtain the relationship curve between the conductivity current density and the applied electric field strength by fitting the test data under different working conditions, so as to identify the electrical aging threshold. This model needs to match the type of film material to which the test sample belongs to ensure the accuracy of the prediction. The relationship prediction model is used to fit the relationship between the conductivity current density and the applied electric field strength, focusing on identifying the electrical aging behavior of the material. The electrical aging threshold of a material refers to the phenomenon that the material begins to show accelerated aging under a certain applied electric field strength. Through this model, the critical point of this accelerated aging can be accurately identified according to the relationship between different temperatures, applied electric field strengths and conductivity current densities, thereby determining the electrical aging threshold.

[0050] Further optionally, first, the conductivity current density data of the test sample is collected at different temperatures and applied electric field strengths to form a training data set. Then, the relationship curve between the conductivity current density and the applied electric field strength is fitted based on the training data set. Next, an inflection point detection algorithm or first-order derivative analysis is used to identify the jump position in the curve, that is, the point where the conductivity current density increases significantly, and the electric field strength corresponding to this position is the electrical aging threshold. Finally, the model is optimized and verified according to the material type to ensure that the model has good predictive ability for the electrical aging behavior of different thin film materials.

[0051] Exemplarily, the relationship prediction model needs to be tuned according to the type of different film materials. Specifically, the model needs to be fine-tuned for different properties of the material (such as dielectric constant, thickness, molecular structure, etc.) to ensure that the fitted relationship curve accurately reflects the electrical aging behavior of the material. In addition, the accuracy of the model can be tested and verified on materials with known electrical aging thresholds. If the model can accurately predict the electrical aging thresholds of these materials, it indicates that the model has high reliability. For example, assuming that a certain polyimide film material is tested, electric fields of different strengths (such as 5 kV / mm, 10 kV / mm, 15 kV / mm) are applied at different temperatures (such as 80°C, 150°C, 200°C) and the electrical conductivity current density is measured. These data are input into the relationship prediction model to fit the relationship curve between the electric field strength and the electrical conductivity current density. Through the inflection point detection algorithm, the jump point of the electrical conductivity current density is identified, for example, a jump occurs when the electric field strength is 13.2 kV / mm. It is determined that 13.2kV / mm is the electrical aging threshold of the polyimide material under specific temperature conditions.

[0052] In an embodiment of the present application, the rate of change of the conductivity current density corresponding to the jump position in the relationship curve is greater than the set change rate threshold. In the fitted relationship curve, the jump position is usually the point where the conductivity current density suddenly increases significantly. This indicates that the material gradually transitions from an insulating state to a conductive state, that is, the critical point at which the material begins to show electrical aging. Further optionally, the slope change of the curve can be analyzed by a first-order derivative, or an inflection point detection algorithm (such as a second-order derivative analysis of curve fitting) can be used to find the position where the conductivity current density changes significantly. The applied electric field strength corresponding to the jump position is the electrical aging threshold. For example, when the curve rises significantly at a certain electric field strength (such as 13 kV / mm), it indicates that the point is the electrical aging threshold of the material.

[0053] As an optional example, a hybrid neural network model is used to optimize the identification of the jump position of the relationship between the conductive current density and the applied electric field strength. For example, the following neural network and algorithm combination is used to ensure accurate detection of the electrical aging threshold, especially to optimize the identification of the jump position.

[0054] First, data on the conductivity current density is collected at different temperatures and applied electric field strengths to form a training data set. The input dimensions of the data set include: temperature, electric field strength, and material properties (dielectric constant, thickness, molecular structure, etc.). Furthermore, the collected conductivity current density data is normalized or standardized so that the neural network can learn data features more effectively. For time series data, smoothing or interpolation can also be performed to reduce the impact of noise on the fitting curve.

[0055] Furthermore, the following neural network model can be used to fit the relationship curve between the conductive current density and the applied electric field strength, and detect the jump position. For example, a hybrid convolutional neural network is used, such as a first basic network, which is responsible for extracting the local characteristics and change trends between the conductive current density and the applied electric field strength. The second basic network is used to process the timing dependence under different temperature and material conditions and capture the electrical aging behavior that changes over time. Further, a deep residual network (Attention-ResNet) enhanced by a self-attention mechanism. ResNet extracts the nonlinear relationship between the electric field strength and the conductive current density, and uses the attention mechanism to make the model focus on the key positions (such as the jump points) of the relationship curve. The model can better handle multi-dimensional inputs (such as temperature, material properties, etc.) and automatically detect the position of the curve where the curve changes dramatically. After using the trained neural network model to fit the relationship curve between the conductive current density and the electric field strength, the slope change of the curve is calculated by performing a first-order derivative analysis on the fitted curve, and the point where the slope increases sharply is identified as the jump position where the conductive current density changes significantly. The second-order derivative is used to detect the change in curvature, find the inflection point in the curve, that is, the location where the slope changes sharply, and further accurately identify the jump point. Optionally, the neural network is used to learn the characteristics of the jump point under different materials and temperature conditions, and the threshold is automatically set. For example, when the rate of change of the conductive current density exceeds a certain adaptive threshold, it is determined to be a jump point. Furthermore, the fitting curve output by the neural network model is combined with the traditional inflection point detection algorithm, making the jump point identification process more robust and able to more accurately detect the threshold of electrical aging.

[0056] What can be further optimized is that the model may perform differently on different thin film materials, so it is necessary to tune the material properties. Specifically, the dielectric constant, thickness, molecular structure and other characteristics of the material are added as additional input features to the input layer of the neural network. In this way, the electrical aging curve of a specific material can be fitted according to the characteristics of different materials, thereby improving the generalization ability of the model. In other words, the model parameters can be dynamically adjusted according to the characteristics of different materials to fit more accurate electrical aging behavior.

[0057] Further optionally, through the multi-task learning mechanism, the model can simultaneously fit the relationship curve between the conductivity current density and the electric field strength and predict the jump position. This allows the model to learn the differences in electrical aging behavior on different thin film materials at the same time. For example, a multi-task learning framework is designed so that the model not only fits the relationship curve between the conductivity current density and the electric field strength, but also predicts the electrical aging threshold of the material at the same time. This helps the model learn more related information under different material properties. For example, a branch network structure is constructed in which each branch processes a set of related features. For example, in the main network, the main processing is temperature, conductivity current density, and electric field strength. The material property network mainly processes material property inputs (dielectric constant, thickness, molecular structure, etc.). Then, the two sets of features are fused in the middle layer of the neural network to better predict the electrical aging threshold. Specifically, the specific logic of the network is as follows: the input layer receives two sets of input features, the first set is electrical properties, including conductivity current density, electric field strength, and temperature. These features pass through a 128-unit fully connected layer (Dense layer) and then through a 64-unit fully connected layer to extract important information. The second group is material properties, including dielectric constant, thickness, and molecular structure. These features also pass through a 64-unit fully connected layer, and then through a 32-unit fully connected layer to extract the key features of material properties. After the electrical properties and material properties are extracted by their respective networks, the two sets of features are fused through the concatenate operation. The fused features contain rich information about both material properties and electrical parameters. The fused features are processed by a 64-unit fully connected layer, and then by a 32-unit fully connected layer to further learn the complex relationship between the conductive current density and the electric field strength. Finally, the information after feature fusion and processing passes through an output layer to predict the jump point (electrical aging threshold) in the relationship between the conductive current density and the electric field strength. The output layer is a linear activation function that returns the predicted value. The model is optimized using an optimizer, and the loss function is the mean square error to ensure that the model can effectively fit and predict the electrical aging threshold. In this way, electrical properties and material properties can be used at the same time to better fit the electrical aging curve through feature fusion and accurately predict the jump point.

[0058] After the model is trained, it needs to be tested and verified with materials with known electrical aging thresholds. You can select some experimentally verified thin film materials to test the prediction accuracy of the model under different temperature and electric field strength conditions. If the model can accurately predict the electrical aging thresholds of these known materials, it means that the model has good reliability. Perform multiple cross-validations on the model to ensure the generalization ability and stability of the model under different data sets. By evaluating the difference between the fitting curve and the real data (such as mean square error, R² and other indicators), continuously optimize the model's hyperparameters and improve the model's accuracy.

[0059] For example, suppose a certain polyimide film material is tested, and electric fields of different strengths (such as 5 kV / mm, 10 kV / mm, 15 kV / mm) are applied at different temperatures (80°C, 150°C, 200°C), and the conductivity current density is measured. Multidimensional features such as temperature, electric field strength, and conductivity current density are input into the neural network. The relationship curve between conductivity current density and electric field strength is fitted using the CNN+LSTM or Attention-ResNet model. The first-order derivative or inflection point detection algorithm is used to identify the jump point of the conductivity current density, for example, when the electric field strength is 13.2 kV / mm, the jump point is determined to be the electrical aging threshold of the polyimide material. The model is fine-tuned according to the dielectric constant, thickness and other characteristics of the material, and verified on materials with known electrical aging thresholds.

[0060] By introducing the hybrid neural network model, combined with the first-order derivative and inflection point detection algorithm, the jump position identification process in the relationship between the conductive current density and the applied electric field strength can be effectively optimized. Tuning and verification are carried out for different material properties to ensure that the model can accurately reflect the electrical aging behavior of different thin film materials and improve the accuracy and reliability of electrical aging threshold prediction.

[0061] Further optionally, the corresponding relationship between the conductive current density and the applied electric field strength is expressed as: . Where J is the electrical current density, is the kth applied electric field strength and A is a constant.

[0062] In 201, different types of thin film materials are selected as test samples to ensure that the materials have obvious electrical aging characteristics under the action of electric fields. Under different temperatures (such as 80℃, 150℃, 200℃, etc.), a series of electric field strengths (such as 5 kV / mm, 10 kV / mm, 15 kV / mm, etc.) are applied to the test samples, and the conductivity current density at each electric field strength is recorded. The conductivity current density (J) and applied electric field strength (E) data at each test temperature are input into the relationship prediction model. Under each temperature condition, a set of data points are generated, which contain the conductivity current density values ​​under different electric field strengths. According to the input data, an appropriate fitting method (such as linear regression, nonlinear regression or machine learning algorithm) is used to fit the relationship curve between the conductivity current density and the applied electric field strength under each temperature condition. If there is a linear relationship between the conductivity current density and the electric field strength, the linear regression method can be used. If the relationship between the two is more complex (such as a nonlinear relationship), nonlinear methods such as polynomial regression and exponential regression, or more advanced machine learning algorithms (such as support vector machine regression, decision tree, etc.) can be used.

[0063] Further optionally, the output result is usually displayed in the form of a curve, with the horizontal axis representing the applied electric field strength and the vertical axis representing the conductive current density. A fitting curve is obtained at each different temperature.

[0064] In 202, in the fitted relationship curve, the jump point of the conductivity current density is found, that is, the transition point where the curve changes from a relatively stable area to a steep rise. This jump position reflects the critical point where the material begins to show obvious electrical aging. The first-order derivative analysis can be to identify the position where the slope of the curve changes sharply by calculating the first-order derivative (slope) of the curve, and the point where the slope suddenly increases is the jump position. Alternatively, an inflection point detection algorithm can also be used, such as finding the inflection point of the curve through second-order derivative analysis, or identifying the jump point by performing local extreme point detection on the fitting curve. After the jump position, the applied electric field intensity corresponding to the point is recorded as the electrical aging threshold of the test sample at the temperature. For example, assuming that at a temperature of 150°C, the relationship curve of conductivity current density jumps significantly at 13 kV / mm, then 13 kV / mm is the electrical aging threshold of the material at 150°C. By analyzing the relationship curves and their jump positions under different temperature conditions, the electrical aging threshold curves of the material at various temperatures can be obtained. Generally, the electrical aging threshold decreases with increasing temperature because the electrical strength of the material may weaken at high temperatures.

[0065] Step 201 and step 202 form a complete set of electrical aging threshold analysis processes based on electrical current density and electric field strength, from data collection, model fitting to jump position identification and then threshold determination. By fitting the test data of different temperatures and electric field strengths and finding the jump point in the curve, the electrical aging critical point of the material can be accurately determined, thereby evaluating the insulation performance and durability of the material.

[0066] In an embodiment of the present application, the multiple test environments respectively use different temperatures and different applied electric field strengths. Further optionally, the applied electric field strengths used in the multiple test environments are all less than the target applied electric field strength corresponding to the electrical aging threshold. Specifically, multiple test environments are set with different temperatures and applied electric field strengths to test the corona resistance of the electromagnetic wire insulation layer. The electrical aging threshold here refers to the critical electric field strength at which the electromagnetic wire insulation material begins to age under a certain temperature and electric field strength. In order to ensure the accuracy and reliability of the test, the applied electric field strength under all test environments is less than this electrical aging threshold to avoid irreversible damage to the material under overly extreme conditions, thereby ensuring that the test results are closer to the material performance in actual applications.

[0067] In the embodiment of the present application, the target sample corresponds to at least one thin film material to be tested. Further optionally, the operating environment parameter at least includes the real-time temperature and the real-time applied electric field strength in the operating environment where the target sample is located.

[0068] In the embodiment of the present application, the target sample corresponds to at least one thin film material to be tested, and the operating environment parameters are used to monitor and analyze the performance of these thin film materials under different working conditions. Specifically, the operating environment parameters at least include the real-time temperature of the target sample and the real-time applied electric field strength, which are the key factors affecting the performance and life of the thin film material.

[0069] Real-time temperature refers to the temperature actually measured by the target sample in its working environment, reflecting the heat load of the sample's environment. The properties of thin film materials (such as mechanical strength, corona resistance, insulation strength, etc.) usually change with temperature. Under high temperature conditions, the thermal aging of the material is aggravated and the insulation performance may decrease. Real-time temperature monitoring can help analyze the performance of the material within a specific temperature range and evaluate its thermal stability and aging resistance. During the operation of the motor or transformer, the real-time temperature will fluctuate due to factors such as the thermal effect of the current and changes in ambient temperature. By monitoring the real-time temperature of the thin film material, you can accurately understand its heating conditions under working conditions.

[0070] Real-time applied electric field strength refers to the electric field strength applied to the film material in real time during operation, usually in volts per millimeter (V / mm). The applied electric field strength directly affects the electrical properties of the film material. Excessive electric field strength may lead to electrical aging and even corona discharge. Real-time electric field strength monitoring helps to determine whether the material is operating within the safe range and evaluate its electrical aging life. In motor windings or transformer coils, the electric field strength to which the insulating film is subjected varies with the operating voltage and load. Through real-time monitoring, it can be ensured that the electric field strength is maintained within the safe operating range of the material (usually below the electrical aging threshold), thereby avoiding premature material aging or failure.

[0071] During the operation of high-performance motors, it is very important to monitor the temperature and electric field strength of thin-film insulation materials in real time. Assume that the electrical aging threshold of a certain material is 15 kV / mm, and its operating temperature range is -40°C to 200°C. Through online monitoring of the motor, you can always know the current operating temperature (e.g. 150°C) and applied electric field strength (e.g. 12 kV / mm) of the material. When these two parameters are always within the safe operating range of the material, the material can maintain its insulation performance for a longer time. If the temperature or electric field strength exceeds the safe range, timely measures need to be taken to reduce the operating load to prevent premature failure of the insulation material.

[0072] In general, real-time temperature and applied electric field strength monitoring plays a key role in the performance evaluation and life prediction of thin film materials. By monitoring these parameters, the aging process of materials can be better predicted and managed to ensure the stable operation of the equipment.

[0073] As an optional embodiment, in 103, an accelerated corona resistance test is performed based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments.

[0074] In this optional embodiment, the accelerated corona resistance test is a material testing method based on the previously determined electrical aging threshold, the purpose of which is to quickly evaluate the material's corona resistance performance and obtain its short-term life under different test environments. This process accelerates the material's electrical aging process by applying an electric field strength exceeding the electrical aging threshold, thereby inferring its performance and life in actual work.

[0075] In the accelerated corona resistance test based on the electrical aging threshold, the test sample must first be selected to ensure that it can undergo electrical aging or corona discharge under the action of the electric field. According to the previous steps (201 and 202), the electrical aging threshold of the test sample is determined, for example, 13 kV / mm. Then, multiple test environments are designed, including different temperatures (such as room temperature, 100°C, 150°C), humidity (such as 40%RH, 60%RH) and electric field strength (usually an electric field strength higher than the threshold is selected, such as 15 kV / mm). In each test environment, an electric field strength exceeding the electrical aging threshold is applied, and the start time and subsequent changes of the corona discharge of the material are recorded in real time by the corona discharge monitoring equipment. The high voltage is continuously applied and monitored until the material shows obvious corona damage, and the short-term corona resistance life is recorded. This process is repeated in different environments, and the life values ​​under each condition are recorded. For example, under the conditions of 15 kV / mm electric field strength and 150°C, the life may be 30 hours, 100 hours at room temperature, and 50 hours under high humidity. Finally, by collating and comparing the life data under different environments, we analyze the effects of temperature, humidity and electric field strength on life. High temperature and high humidity usually accelerate material aging and shorten life. Based on the test results, we can establish an electrical aging model for the material to predict its performance in long-term use, which provides a basis for material selection and product optimization.

[0076] As an optional embodiment, in 103, after performing an accelerated corona resistance test based on the electrical aging threshold and obtaining the short-time corona resistance life test value of the test sample under multiple test environments, at least two groups of original test values ​​can be selected from the short-time corona resistance life test values ​​of the test sample under multiple test environments. Among them, the original test values ​​at least include: the short-time corona resistance life of the test sample under multiple test environments, the number of damages per unit time, the failure time, the electric field strength, and the temperature. Then, multiple groups of random weight parameters are generated, and the multiple groups of random weight parameters are used to perform linear interpolation mixing on at least two groups of original test values ​​respectively to obtain corresponding multiple groups of first mixed data. Then, the multiple groups of first mixed data are input into the environmental simulation model, and outlier screening and numerical optimization are performed based on the attribute characteristics corresponding to the test environment to obtain multiple groups of second mixed data. Then, the multiple groups of second mixed data are fused with the original test values ​​to obtain the optimized short-time corona resistance life test value.

[0077] In the embodiment of the present application, the electric field strength used for testing in the test environment is less than the electrical aging threshold, and the electric field strength used for testing in the test environment is greater than the actual applied electric field strength in the actual environment.

[0078] For example, after performing accelerated corona resistance tests under multiple test environments, the original test data is obtained. The original test values ​​here include the following items:

[0079] Short-time corona resistance life: The corona resistance life of the test sample under each test environment, such as 30 hours, 50 hours, 100 hours, etc.

[0080] Number of destructions per unit time: The number of material destruction or corona discharges that occur per unit time during the test.

[0081] Failure time: The point in time when irreversible damage occurs to the material.

[0082] Electric field strength: The electric field strength applied to the test sample, usually between the electrical aging threshold and the actual working electric field.

[0083] Temperature: The temperature conditions in each test environment, such as room temperature, 100°C, 150°C, etc.

[0084] Based on the above assumptions, at least two groups of test values ​​(such as test data under different temperature conditions) are selected from the original test values, and then multiple groups of random weight parameters are generated. These weights are used to perform linear interpolation mixing on at least two groups of original test values ​​to generate a new set of data (called the first mixed data). The purpose of linear interpolation mixing is to generate new data points between the original data to help simulate more environmental conditions and obtain a more complete data set. For example, assuming that under an electric field strength of 15 kV / mm, the short-time corona resistance life in an environment of 150°C is 30 hours and 100 hours at room temperature, the random weight parameters can be interpolated between the two to generate a new life data, such as 65 hours.

[0085] Then, the generated multiple sets of first mixed data are input into the environmental simulation model. The environmental simulation model processes these data by considering the attribute characteristics of various test environments (such as temperature, electric field strength, humidity, etc.), screens out abnormal values ​​(which may be the part of the data that deviates from the reasonable range), and numerically optimizes the data based on the test environment conditions to obtain a more reasonable set of optimized data (called second mixed data). For example, if a certain interpolated data point deviates significantly from the test trend, the model may determine it as an abnormal value and adjust it to make it more consistent with the actual corona resistance performance of the material.

[0086] Next, multiple sets of second mixed data are fused with the original test values ​​to finally obtain the optimized short-time corona resistance life test value. This fusion process can help further improve the accuracy of the test results, eliminate the deviation caused by individual test conditions or experimental errors, and obtain a more stable and reliable material corona resistance life prediction value.

[0087] In this assumption, the original test data is known as: under the electric field strength of 15 kV / mm, the short-time corona resistance life at 150℃ is 30 hours, the failure time is 29 hours, and the number of damages per unit time is 3 times / hour; the short-time corona resistance life at room temperature is 100 hours, the failure time is 98 hours, and the number of damages per unit time is 1 time / hour. By generating random weights (such as 0.6 and 0.4), these two sets of data are linearly interpolated and mixed to obtain the interpolated data: the short-time corona resistance life is 65 hours, the failure time is 57.8 hours, and the number of damages per unit time is 2.2 times / hour. Then, this set of mixed data is input into the environmental simulation model for outlier screening and adjustment. For example, if the number of damages per unit time after interpolation is too high, the model will identify it as an outlier and adjust it for optimization. Finally, after the data is optimized, the short-time corona resistance life is adjusted to 60 hours, the failure time is 56 hours, and the number of damages per unit time is 2 times / hour. These optimized data are then fused with other original test data to ultimately obtain a more accurate prediction of the short-time corona resistance and corona resistance life.

[0088] In summary, the accuracy of the short-term corona resistance life test value can be effectively improved by linearly interpolating and mixing the original test values ​​under multiple test environments, inputting the environmental simulation model for abnormal screening and numerical optimization, and then fusing the optimized data with the original test values. This process not only considers the relationship between multiple groups of data, but also simulates and optimizes the abnormal values, making the final result more valuable for reference.

[0089] As an optional embodiment, generating multiple groups of random weight parameters in the above steps can be implemented as follows: according to the real-time environmental parameters of the test environment, selecting the data distribution type that matches the test sample under multiple test environments from the original data distribution library; generating candidate random weight parameters that match the respective distribution characteristics based on the multiple data distribution types; simulating the change trend of the candidate random weight parameters according to the thin film material properties corresponding to the test sample; and making a rationality judgment based on the simulation results to obtain the random weight parameters that are finally output.

[0090] In an embodiment of the present application, the random weight parameters at least include: distribution shape parameters, weighted ratio parameters, random strategy selection parameters, and mixing ratio control parameters. In an embodiment of the present application, the random weight parameters are used to perform linear interpolation mixing on the original data under different test conditions to generate a richer and more accurate test data set. The distribution shape parameters are used to control the distribution shape when generating random weights, such as normal distribution, uniform distribution, etc. The choice of distribution shape will affect the way the weights are generated, thereby affecting the interpolation results of the data. Selecting different distribution shapes can simulate different random change patterns. For example, when using a normal distribution, the weights will fluctuate around a central value, while when using a uniform distribution, the weights will fluctuate over a wider range and the generated data will be more dispersed. Exemplarily, under a normal distribution, the weights may be more concentrated near the middle value; under a uniform distribution, the generated weights are more evenly distributed between the maximum and minimum values.

[0091] The weighting ratio parameter determines the weight ratio of different raw data in the linear interpolation mixture. For example, when the weighting parameters are 0.6 and 0.4, it means that in the interpolation process, the data at 150℃ and room temperature account for 60% and 40% respectively. By controlling the weighting ratio, the contribution of two or more groups of data can be adjusted, thereby affecting the bias of the generated new data. The larger the weighting ratio, the greater the influence of the raw data on the final result. For example, if the weighting ratio is 0.7 and 0.3, the interpolation result will be closer to the data at 150℃.

[0092] The random strategy selection parameter is used to control how to select the random strategy for weight generation, such as a deterministic strategy or a random strategy. The random strategy may include different interpolation methods, interpolation ranges, or select different data combinations for mixing. By selecting different strategies, the interpolation process can produce diverse results, simulate various possible environmental changes, and enhance the generalization ability of the model. For example, the random strategy may include selecting different combinations of temperature conditions for interpolation, or generating new test values ​​through data under different electric field strengths.

[0093] The mixing ratio control parameter is used to control the specific ratio of linear interpolation mixing, that is, to control the degree of mixing between each group of original data. This parameter determines the balance point between different data, thereby affecting the distribution characteristics of the generated new data. By adjusting the mixing ratio, more flexible interpolation results can be generated. For example, the interpolation result can be controlled to be more biased towards the original data under a specific environment or to be balanced. For example, if the mixing ratio control parameter is set to make the interpolation result more biased towards the data under high temperature conditions, the interpolated test data will be closer to the result under 150℃ environment.

[0094] These random weight parameters work together to determine the weight and influence of different test environment data in the interpolation mixture, thereby generating a richer test data set. The distribution shape parameter determines the distribution form of the weight, the weighted ratio parameter controls the contribution of different data, the random strategy selection parameter determines the way the weight is generated, and the mixing ratio control parameter adjusts the balance of different original data. The reasonable configuration of these parameters helps to generate optimized data that conforms to the actual situation, improve the accuracy of the test results and the reliability of the model prediction.

[0095] Based on the above parameters, in this optional embodiment, the step of generating multiple sets of random weight parameters is achieved through a series of intelligent selection and simulation processes to ensure that the generated weight parameters better reflect the characteristics of the test environment and material properties. Specifically, according to the real-time environmental parameters of the test environment, the data distribution type that matches the test sample under multiple test environments is selected from the original data distribution library.

[0096] Real-time environmental parameters refer to the environmental conditions during the current test, such as temperature, electric field strength, humidity, etc. These conditions will affect the performance of the test sample, so it is necessary to select the original data distribution that matches these parameters. The original data distribution library contains data such as corona resistance life, failure time, number of damages, etc. collected under different test environments. Each type of environment corresponds to a different data distribution type, such as normal distribution, lognormal distribution, uniform distribution, etc. According to the real-time test environment, selecting a distribution type suitable for the current test environment from the data distribution library can ensure that the generated weight parameters are more consistent with the actual test situation and increase the authenticity and accuracy of data interpolation. For example, if the current test environment is high temperature (150°C) and the electric field strength is 15 kV / mm, the system will automatically select a distribution type that matches this condition (for example, normal distribution) to generate subsequent weight parameters.

[0097] Different distribution types have different characteristics, so the generated random weight parameters must match each distribution type. For example, if the normal distribution is selected, the random weight parameters should be distributed in a bell-shaped curve around the mean, with the majority of weights concentrated in the middle. If the uniform distribution is selected, the weight parameters should be evenly distributed within a certain range, with all values ​​having similar probabilities of occurring.

[0098] By generating candidate random weight parameters that conform to the distribution characteristics, the characteristics of each test environment can be better reflected, so that the interpolated data has a stronger representativeness. For example, suppose the candidate random weights generated by the system are 0.4, 0.5, and 0.6. These weights all conform to the characteristics of the normal distribution, that is, most of the weights are concentrated near the middle value.

[0099] Different thin film materials behave differently in different environments, so the generated candidate random weight parameters also need to consider the influence of material properties. Thin film material properties include electrical properties, heat resistance, aging resistance, etc. Based on these properties, the generated candidate random weight parameters are simulated to predict the possible performance of the material in various environments, so as to adjust the weight parameters. By simulating the actual performance of the material, the candidate random weight parameters can be adjusted more accurately to ensure that these parameters are consistent with the characteristics of the material and reflect the influence of environmental conditions. For example, a certain thin film material will age faster in a high temperature environment. Therefore, the weight parameters under high temperature conditions may be reduced during the simulation process to reflect the trend of shortening the material life.

[0100] After simulating the trend of change, the system makes a rationality judgment on the generated random weight parameters to ensure that these weight parameters conform to the actual physical laws and test environment conditions. The basis for judgment includes the consistency of test data, the rationality of data distribution, and the physical reality of material performance in different environments. Rationality judgment can be made through statistical analysis, error testing, etc. If some weight parameters deviate from the reasonable range, the system will adjust them. The final random weight parameters have been simulated and rationally verified for material properties to ensure that they can be accurately used for subsequent data interpolation and mixing to obtain more realistic test data. For example, assuming that the simulation results of a set of weight parameters at high temperature do not match the actual material performance, the system will make adjustments, and the final output weights may be 0.3, 0.4, and 0.5 to ensure that the generated data is reasonable and reliable.

[0101] This embodiment selects data distribution through real-time environmental parameters, generates candidate random weights that meet the distribution characteristics, simulates and adjusts according to material properties, and makes reasonable judgments on the results to ensure that the generated random weight parameters can accurately reflect the characteristics of the test environment and materials. This process greatly improves the reliability and accuracy of the final generated data, providing a more accurate basis for corona resistance life prediction.

[0102] As an optional embodiment, in the above steps, before fusing the multiple sets of second mixed data with the original test values ​​to obtain the optimized short-time corona resistance life test value, multiple original noise types can also be determined and combined into multiple sets of mixed noise types. Furthermore, based on the film material type corresponding to the test sample and the real-time operating parameters under the test environment, the noise intensity, fusion weight and injected data density corresponding to the test sample are determined. Finally, based on the noise intensity, fusion weight and injected data density corresponding to the test sample, the noise data corresponding to each of the multiple sets of mixed noise types are injected into the multiple sets of second mixed data and the original test values ​​to obtain multiple sets of third mixed data.

[0103] In this optional embodiment, the prediction process of short-time corona resistance life is further optimized by introducing and processing multiple original noise types. Original noise types refer to interference signals that are naturally generated and inevitable during the test process or equipment operation. These noises may come from different sources, such as electrical noise, temperature fluctuations, equipment errors, etc. The system determines multiple noise types based on different test conditions (such as temperature, electric field strength, material properties, etc.) and combines them into multiple groups of mixed noise types. These noise types may be independent of each other or have some correlation. Combining multiple noise types can simulate more complex noise environments and ensure that these influencing factors can be fully considered during data interpolation and fusion.

[0104] Exemplarily, electric field fluctuation noise, temperature fluctuation noise, and equipment measurement error noise may be identified, and then these noises may be combined into a group of mixed noise types for subsequent data fusion and noise injection. Noise intensity refers to the proportion or degree of influence of noise in the entire test process. Different test environments and thin film materials have different sensitivities to noise, so it is necessary to determine the appropriate noise intensity for each test sample. For example, the noise intensity in a high temperature environment may be higher than that in a room temperature environment. Fusion weight, when performing data interpolation or data fusion, the influence of each group of noise types needs to be set with a corresponding weight to determine the contribution of noise in generating new data. Injection data density refers to the injection ratio or frequency of noise data in mixed data. The determination of data density should match the noise intensity and weight to ensure that the noise distribution is reasonable and does not cause data distortion. In this way, determining the noise intensity, fusion weight, and injection data density can help the system more accurately simulate the impact of noise on test data and ensure that the generated data is realistic enough.

[0105] If the thin film material is sensitive to noise at high electric fields, the system may set a higher noise intensity for that noise type (e.g. 30%), while a lower intensity may be set for temperature fluctuation noise (e.g. 10%). The fusion weights and implanted data density are also adjusted accordingly to reflect the different impacts of these noises.

[0106] In the process of fusing the generated multiple sets of second mixed data (i.e. the initial data obtained by linear interpolation) and the original test values, the system will inject multiple sets of mixed noise type data according to the noise intensity, fusion weight and injected data density determined in the previous step. The process of noise data injection simulates the noise interference in the real test environment, making the mixed data more realistic and anti-interference.

[0107] The third hybrid data generated after injecting noise data can better reflect the complex noise environment that may be encountered under actual test conditions. By considering the impact of noise on the data, the final optimized data is more robust and accurate, and can be better used for life prediction.

[0108] In high temperature and high electric field environments, the system will inject corresponding noise data, such as electric field fluctuation noise and temperature fluctuation noise. These noise data are gradually injected according to the previously set intensity, weight and density, making the third hybrid data closer to the noise characteristics in the actual operating environment.

[0109] This optional embodiment further optimizes the generation process of the short-time corona resistance and corona resistance life prediction value by introducing and injecting noise data. Specifically, first determine a plurality of original noise types and combine them into a mixed noise type, then determine the noise intensity, fusion weight and injection data density based on the material type and test environment conditions, and finally inject these noise data into multiple sets of second mixed data and original test values ​​to generate third mixed data that is closer to the actual operating conditions. This can effectively improve the reliability and noise interference resistance of the data model, and ultimately obtain a more accurate corona resistance life prediction value.

[0110] As an optional embodiment, constructing an accelerated prediction model based on the short-time corona resistance life test value can be implemented as follows:

[0111] An accelerated prediction model training set is constructed using the short-time corona resistance life test value and the environmental parameters corresponding to the test environment; a life prediction formula is obtained based on the short-time corona resistance life test value; the life prediction formula is used as a constraint generation condition for the physical loss layer to generate corresponding dynamic constraints; based on the dynamic constraints, an initial prediction model is trained in the accelerated prediction model training set to obtain the accelerated prediction model.

[0112] In this optional embodiment, the process of constructing an accelerated prediction model based on short-time corona resistance life test values ​​is mainly divided into four key steps: constructing an accelerated prediction model training set, fitting a life prediction formula, generating dynamic constraints, and training an initial prediction model.

[0113] The short-time corona resistance life test value refers to the optimized test data generated by noise injection, data interpolation and other processes. These data are more realistic and reliable. The environmental parameters corresponding to the test environment include important variables that affect the life of the material, such as temperature, electric field strength, and humidity. These parameters, together with the life test values, constitute the training data. The training set construction process collects the short-time corona resistance life test values ​​and corresponding environmental parameters under different environments, and uses them as input data to form the basic data set of the prediction model. By matching the test values ​​and environmental parameters, a more representative training set of the accelerated prediction model can be generated, providing a rich data foundation for the accurate training of subsequent models.

[0114] Exemplarily, the system collects short-time corona resistance life test values ​​at different temperatures (such as 150°C, 200°C) and different electric field strengths (such as 10 kV / mm, 15kV / mm), and combines these data with the corresponding environmental parameters to form a training set.

[0115] The life prediction formula is obtained by fitting the collected short-time corona resistance life test values. The fitting process utilizes the relationship between data and environmental parameters, and may use regression analysis, nonlinear fitting and other methods. The life prediction formula can describe the relationship between the corona resistance life of the material and environmental factors, usually in a formalized way, such as exponential relationship, power function relationship, etc. This formula can reflect the life change trend of the material under different environments. Through the life prediction formula, a mathematical expression for predicting the life of the material under various environmental conditions can be preliminarily obtained. The formula provides physical constraints to ensure that the subsequent prediction model is consistent with the actual physical phenomenon.

[0116] Further optionally, the life prediction formula is expressed as: Wherein L is the short-time corona resistance life test value, is the electric field strength under the nth test environment, M is a constant, and T is the temperature.

[0117] The physical loss layer refers to using physical laws as constraints on the deep learning model to ensure that the model output follows basic physical laws. The life prediction formula here is the physical constraint.

[0118] Dynamic constraints introduce an adaptive dynamic adjustment mechanism to the life prediction formula, allowing the model to adjust to different environments and material properties during training. These conditions are not limited to conventional error minimization, but also include constraints on the physical rationality of the model output. The generated dynamic constraints prevent the model from deviating from the laws of physics during the learning process, avoiding the model being based solely on data-driven results, but instead combining the physical laws of the life prediction formula to ensure the interpretability and accuracy of the prediction results.

[0119] For example, assuming that the prediction results generated by the model during training deviate from the reasonable range defined by the life prediction formula, the dynamic constraints will dynamically adjust the model weights according to the deviation to ensure that the output results are consistent with the laws of physics.

[0120] The initial prediction model is a model without physical constraints, and is initially trained based on the accelerated prediction model training set. The model forms prediction capabilities by learning the relationship between short-term corona resistance life test values ​​and environmental parameters.

[0121] The accelerated prediction model is to further train and optimize the initial prediction model by introducing dynamic constraints. At this time, the model can not only predict the material life, but also ensure that the prediction results conform to the laws of physics and have good generalization capabilities under different environments.

[0122] Finally, the accelerated prediction model optimized by dynamic constraints can more accurately predict the corona resistance life of materials under different environments. Compared with the traditional model, this model not only takes into account the historical test data, but also introduces physical constraints and noise effects, which improves the reliability of the prediction results.

[0123] For example, suppose the initial model predicts that the material life is 1000 hours, but according to the life prediction formula and environmental parameters, the reasonable life should be 800 hours. After the optimization of dynamic constraints, the predicted life output by the final acceleration prediction model is close to 800 hours, which is in line with the laws of physics.

[0124] This optional embodiment constructs an accelerated prediction model through the following steps: first, construct a training set containing short-time corona resistance life test values ​​and environmental parameters; then obtain a life prediction formula through fitting; generate dynamic constraints based on the formula, and use these conditions in the training set to optimize the initial prediction model. Ultimately, the obtained accelerated prediction model can not only accurately predict the corona resistance life of the material, but also ensure that the output of the model conforms to the actual physical laws, and has better generalization ability and reliability.

[0125] In the embodiment of the present application, further, the accelerated prediction model is used to predict the corona resistance life value at a temperature and electric field strength corresponding to a threshold value of electrical aging. In the embodiment of the present application, an important use of the accelerated prediction model is to predict the corona resistance life value at a temperature and electric field strength corresponding to a threshold value of electrical aging. The electrical aging threshold refers to the critical point at which the material begins to show electrical aging phenomena at a specific temperature and electric field strength. Above this threshold, the electric field and thermal stress inside the material gradually accumulate, resulting in accelerated electrical aging and a significant shortening of the material failure time. Below this threshold, the material is still in a relatively safe working state with a low aging rate. The electrical aging threshold is usually determined experimentally, and it is closely related to the material type, film thickness, and environmental conditions (such as temperature and humidity).

[0126] Determining the electrical aging threshold helps to clarify the safe working range of the material, so that the accelerated prediction model can be used to predict the corona resistance life of the material below the threshold. For example, the electrical aging threshold of a certain film material is 200℃ and 20 kV / mm, which means that above this condition, the electrical aging phenomenon will accelerate. When the temperature is below 200℃ and the electric field strength is below 20 kV / mm, the material life is extended and the aging process is relatively slow.

[0127] In an environment below the electrical aging threshold (i.e., the temperature and electric field strength are below the threshold), the aging rate of the material slows down significantly, but it is still affected by the electric field and thermal stress. Life prediction at this time requires accurate calculation of the long-term performance of the material under lower stress. The accelerated prediction model can predict the corona resistance life of the material under conditions below the electrical aging threshold by learning and fitting historical data, environmental parameters, and physical laws.

[0128] Accelerated prediction models can provide accurate material durability assessments, especially under long-term low-load conditions. This not only helps with life management in engineering applications, but also guides equipment to operate within a safe range to avoid early failures caused by electrical aging. For example, assuming that the electrical aging threshold of the material is a temperature of 200°C and an electric field of 20 kV / mm, an accelerated prediction model is used to predict the material life under conditions of 150°C and 15 kV / mm. Through the calculation of the model, it may be predicted that the corona resistance life of the material under these conditions is 5000 hours, while the life under conditions close to the threshold may be only 2000 hours.

[0129] The accelerated prediction model combines a variety of environmental factors (such as temperature, electric field strength) and short-term corona resistance life test values, and can predict the life of materials under different conditions through the life prediction formula obtained by fitting. When the model predicts the life under conditions less than the electrical aging threshold, it can ensure that the output life value conforms to the actual physical laws through dynamic constraints. The model is not only accurate for extreme conditions exceeding the threshold, but also can provide accurate prediction results under conditions below the threshold.

[0130] This prediction method based on the accelerated prediction model makes the prediction of material life under low stress conditions more reliable, especially below the electrical aging threshold, and can give a long-term durability prediction of the material, which is beneficial to the long-term safe operation of materials and equipment.

[0131] The accelerated prediction model in the embodiment of the present application can effectively predict the corona resistance life of materials at temperatures and electric field strengths less than the electrical aging threshold by using short-term corona resistance life test values ​​and environmental parameters combined with physical constraints. This is crucial to ensure the long-term operation of materials under safe conditions, helping to extend the life of materials and reduce the risk of failure.

[0132] As an optional embodiment, the lifetime prediction formula is used as the constraint generation condition of the physical loss layer to generate the corresponding dynamic constraint condition, which can be implemented as follows:

[0133] The corresponding relationship between the conductive current density and the applied electric field strength and the life prediction formula are mixed to generate the corresponding dynamic constraint condition; the dynamic constraint condition is expressed as the following formula:

[0134] in, is the value of the physical loss function, is the predicted value of corona resistance life, is the i-th conductance current density, is the ith applied electric field strength in the nth test environment, A is a constant, is the total number of test environments.

[0135] In this optional embodiment, dynamic constraints are generated by combining the corresponding relationship between the conductive current density and the applied electric field strength with the life prediction formula. The specific formula describes how the physical loss function introduces the relationship between the two to ensure the physical consistency of the prediction model. Dynamic constraints are generated by combining the relationship between the conductive current density and the applied electric field strength and the life prediction formula. The physical loss function measures the deviation between the model prediction value and the physical law, and is used to optimize the model during the training process to make it more accurately predict the corona resistance life. This method ensures that the prediction model is physically reasonable and can be well adapted to test conditions under different environmental conditions.

[0136] In the embodiment of the present application, first, the method collects data of the conductive current density and the applied electric field strength at different temperatures, and determines the electrical aging threshold through these data. The electrical aging threshold is the critical point at which the material begins to age significantly under certain electric field and temperature conditions. Based on this threshold, the electric field strength can be selected in the accelerated test in a targeted manner so that the test is carried out without exceeding the electrical aging threshold, thereby accelerating the life test and ensuring the validity and relevance of the test results. By performing accelerated corona resistance tests at different temperatures and different applied electric field strengths, the method can quickly obtain the short-term corona resistance life of the test sample. The applied electric field strengths of multiple test environments are all less than the electrical aging threshold, so that the test can simulate the operating environment in actual use, while avoiding unrealistic damage caused by excessively intense test conditions. The test data under such multi-environment conditions provides rich sample information, which helps to improve the accuracy and generalization ability of the subsequent accelerated prediction model. Based on the above-mentioned short-term corona resistance life test value, the method constructs an accelerated prediction model. The model uses multiple sets of experimental data and optimizes the neural network in combination with physical constraints (such as the aforementioned physical formula constraints), so that the model can not only fit the test data, but also follow the physical laws of electrical aging and corona resistance life. By introducing physical constraints, the model prediction results can be more reliable, avoiding overfitting experimental data, while ensuring that the model's prediction ability remains consistent under different environmental conditions. The accelerated prediction model can be used for different types of film materials, not just limited to the prediction of the corona resistance life of a specific material. The model input includes real-time operating environment parameters such as temperature and applied electric field strength, which allows the model to make dynamic predictions based on actual operating conditions. This is very useful for predicting the life of electromagnetic wire insulation under different environments and can be applied to a variety of industrial application scenarios. This method can quickly obtain the corona resistance life prediction results of the target sample by inputting the operating environment parameters (such as temperature and electric field strength) of the target sample into the accelerated prediction model. This provides an effective tool for the life assessment of electromagnetic wire insulation materials in practical applications, which can help engineers judge the durability of the material in advance and make appropriate maintenance or replacement plans.

[0137] In summary, by combining physical models and short-term experimental data, the accelerated prediction model can provide more accurate life prediction results, avoiding the errors that may be caused by relying solely on experimental data. The accelerated test method can quickly obtain relevant data on corona resistance life, reducing the need for long-term testing under real operating conditions. It should be noted that this method is not only applicable to a certain specific material, but can also be extended to different types of thin film materials, with good generalization ability. The accelerated prediction model used in this method can be dynamically predicted according to real-time operating conditions (temperature, electric field strength, etc.), and has strong practicality. This method provides an efficient and accurate means of predicting the corona resistance life of electromagnetic wire insulation layer by combining experiments with physical constraints. Through the accelerated prediction model, the corona resistance life of the target sample can be predicted in real time according to the operating environment parameters of the target sample, thus providing strong support for material management in industrial applications.

[0138] In another embodiment of the present application, a system for accelerating the prediction of the corona resistance life of the insulation layer of an electromagnetic wire is provided, which is applied to the prediction of the corona resistance life of thin film materials. Figure 3 The system for accelerating the prediction of the corona resistance life of the electromagnetic wire insulation layer comprises:

[0139] A collection unit is configured to collect the electrical conduction current density and applied electric field strength of a test sample in the electromagnetic wire insulation layer at different temperatures; the test sample corresponds to at least one thin film material to be tested;

[0140] a determination unit configured to determine an electrical aging threshold corresponding to the test sample based on the electrical conduction current density and the applied electric field strength at different temperatures;

[0141] A testing unit is configured to perform an accelerated corona resistance test based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments; wherein the multiple test environments respectively use different temperatures and different applied electric field strengths; and the applied electric field strengths used in the multiple test environments are all less than the target applied electric field strength corresponding to the electrical aging threshold;

[0142] A construction unit configured to construct an accelerated prediction model based on the short-time corona resistance life test value;

[0143] The prediction unit is configured to input the operating environment parameters of the target sample into the accelerated prediction model to obtain the target corona resistance life corresponding to the target sample; wherein the target sample corresponds to at least one thin film material to be tested; and the operating environment parameters at least include the real-time temperature and the real-time applied electric field strength in the operating environment of the target sample.

[0144] Further optionally, the determining unit determines the electrical aging threshold corresponding to the test sample based on the electrical current density and the applied electric field strength at different temperatures, and is configured as follows:

[0145] Input the conductivity current density and applied electric field strength of the test sample at different temperatures into the relationship prediction model, and fit the relationship curve between the conductivity current density and applied electric field strength corresponding to the test sample; the relationship prediction model matches the type of thin film material to which the test sample belongs;

[0146] The corresponding relationship between the conductive current density and the applied electric field strength is expressed as: ; where J is the electrical current density, is the kth applied electric field strength, A is a constant;

[0147] The jump position in the relationship curve is identified, and the applied electric field strength corresponding to the jump position is used as the electrical aging threshold; the change rate of the conductive current density corresponding to the jump position in the relationship curve is greater than the set change rate threshold.

[0148] Further optionally, the testing unit, after performing an accelerated corona resistance test based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments, is further configured to:

[0149] At least two groups of original test values ​​are selected from the short-time corona resistance life test values ​​of the test sample under multiple test environments; wherein the original test values ​​at least include: the short-time corona resistance life, the number of damages per unit time, the failure time, the electric field strength, and the temperature of the test sample under multiple test environments;

[0150] Generate multiple sets of random weight parameters, and use the multiple sets of random weight parameters to perform linear interpolation mixing on at least two sets of original test values, respectively, to obtain corresponding multiple sets of first mixed data;

[0151] Inputting multiple sets of first mixed data into the environment simulation model, performing outlier screening and numerical optimization based on the attribute characteristics corresponding to the test environment, and obtaining multiple sets of second mixed data;

[0152] The multiple sets of second mixed data are combined with the original test values ​​to obtain optimized short-time corona resistance life test values.

[0153] Further optionally, the test unit generates multiple sets of random weight parameters and is configured to:

[0154] According to the real-time environment parameters of the test environment, selecting the data distribution type that matches the test sample in multiple test environments from the original data distribution library;

[0155] Generate candidate random weight parameters matching respective distribution characteristics based on multiple data distribution types;

[0156] Simulating the variation trend of the candidate random weight parameters according to the film material properties corresponding to the test sample;

[0157] A rationality judgment is made based on the simulation results to obtain the random weight parameters of the final output; wherein the random weight parameters include at least: distribution shape parameters, weighted ratio parameters, random strategy selection parameters, and mixed ratio control parameters.

[0158] Further optionally, before the test unit fuses the multiple sets of second mixed data with the original test value to obtain the optimized short-time corona resistance life test value, it is further configured as follows:

[0159] Determine a plurality of original noise types, and combine the plurality of original noise types into a plurality of groups of mixed noise types;

[0160] Determine the noise intensity, fusion weight and injection data density corresponding to the test sample based on the type of thin film material corresponding to the test sample and the real-time operating parameters under the test environment;

[0161] Based on the noise intensity, fusion weight and injected data density corresponding to the test sample, multiple groups of noise data corresponding to the respective mixed noise types are injected into the multiple groups of second mixed data and the original test values ​​to obtain multiple groups of third mixed data.

[0162] Further optionally, the construction unit constructs an accelerated prediction model based on the short-time corona resistance life test value, and is configured to:

[0163] Using the short-time corona resistance life test value and the environmental parameters corresponding to the test environment, constructing an accelerated prediction model training set;

[0164] The life prediction formula is obtained based on the short-time corona resistance life test value fitting; the life prediction formula is expressed as: ; Wherein L is the short-time corona resistance life test value, is the electric field strength under the nth test environment, M is a constant, and T is the temperature;

[0165] Using the life prediction formula as a constraint generation condition for the physical loss layer, generating corresponding dynamic constraint conditions;

[0166] Based on the dynamic constraint condition, an initial prediction model is trained in the accelerated prediction model training set to obtain the accelerated prediction model;

[0167] The accelerated prediction model is used to predict the corona resistance life value at a temperature and electric field strength less than the electrical aging threshold.

[0168] Further optionally, the construction unit uses the life prediction formula as the constraint generation condition of the physical loss layer to generate the corresponding dynamic constraint condition, and is configured as follows:

[0169] The corresponding relationship between the conductive current density and the applied electric field strength and the life prediction formula are mixed to generate the corresponding dynamic constraint condition; the dynamic constraint condition is expressed as the following formula:

[0170]

[0171] in, is the value of the physical loss function, is the predicted value of corona resistance life, is the i-th conductance current density, is the ith applied electric field strength in the nth test environment, A is a constant, is the total number of test environments.

[0172] Further optionally, the strength of the electric field used for testing in the test environment is less than the electrical aging threshold, and the strength of the electric field used for testing in the test environment is greater than the strength of the actually applied electric field in the actual environment.

[0173] In an embodiment of the present application, this accelerated prediction system for the corona resistance life of the electromagnetic wire insulation layer constructs an accelerated prediction model through short-time corona resistance life test values, and uses the accelerated prediction model to predict the corresponding corona resistance life in real time, thereby improving the efficiency of the accelerated prediction of the corona resistance life of the electromagnetic wire insulation layer and improving the accuracy of the corona resistance life prediction value.

[0174] In another embodiment of the present application, a device for accelerating the prediction of the corona resistance life of the insulation layer of an electromagnetic wire is also provided. Figure 4 As shown, it includes: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; the memory is used to store computer programs; the processor is used to implement the method for accelerating the prediction of the corona resistance life of the electromagnetic wire insulation layer described in the method embodiment when executing the program stored in the memory. The communication bus 1140 mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus 1140 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4It is represented by only one thick line, but it does not mean that there is only one bus or one type of bus. The communication interface 1120 is used for communication between the above-mentioned electronic device and other devices. The memory 1130 may include a random access memory (Random Access Memory, RAM), and may also include a non-volatile memory (non-volatile ememory), such as at least one disk storage. Optionally, the memory may also be at least one storage device located away from the aforementioned processor. The above-mentioned processor 1110 may be various types of special-purpose processors. Accordingly, an embodiment of the present application also provides a computer-readable storage medium storing a computer program, which, when executed, can implement the steps that can be performed by the electronic device in the above-mentioned method embodiment.

Claims

1. A method for accelerating the prediction of the corona resistance life of an electromagnetic wire insulation layer, characterized in that: Applied to the scenario of corona resistance life prediction of thin film materials, the method comprises: For a test sample in an electromagnetic wire insulation layer, collecting the electrical conduction current density and applied electric field strength of the test sample at different temperatures; the test sample corresponds to at least one type of thin film material; Determining the electrical aging threshold corresponding to the test sample based on the electrical conduction current density and the applied electric field strength at different temperatures; Performing an accelerated corona resistance test based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments; wherein the multiple test environments respectively use different temperatures and different applied electric field strengths; the applied electric field strengths used in the multiple test environments are all less than the target applied electric field strength corresponding to the electrical aging threshold; Constructing an accelerated prediction model based on the short-time corona resistance life test value; The operating environment parameters of the target sample are input into the accelerated prediction model to obtain the target corona resistance life corresponding to the target sample; wherein the target sample corresponds to at least one thin film material to be tested; and the operating environment parameters include at least the real-time temperature and the real-time applied electric field strength in the operating environment of the target sample.

2. The method for accelerating the prediction of the corona resistance life of the electromagnetic wire insulation layer according to claim 1, characterized in that: The step of determining the electrical aging threshold corresponding to the test sample based on the electrical current density and the applied electric field strength at different temperatures includes: Input the conductivity current density and applied electric field strength of the test sample at different temperatures into the relationship prediction model, and fit the relationship curve between the conductivity current density and applied electric field strength corresponding to the test sample; the relationship prediction model matches the type of thin film material to which the test sample belongs; The corresponding relationship between the conductive current density and the applied electric field strength is expressed as: ; where J is the electrical current density, is the kth applied electric field strength, A is a constant; The jump position in the relationship curve is identified, and the applied electric field strength corresponding to the jump position is used as the electrical aging threshold; the change rate of the conductive current density corresponding to the jump position in the relationship curve is greater than the set change rate threshold.

3. The method for accelerating prediction of corona resistance life of electromagnetic wire insulation layer according to claim 1, characterized in that: After performing the accelerated corona resistance test based on the electrical aging threshold to obtain the short-time corona resistance life test value of the test sample under multiple test environments, the method further includes: At least two groups of original test values ​​are selected from the short-time corona resistance life test values ​​of the test sample under multiple test environments; wherein the original test values ​​at least include: the short-time corona resistance life, the number of damages per unit time, the failure time, the electric field strength, and the temperature of the test sample under multiple test environments; Generate multiple sets of random weight parameters, and use the multiple sets of random weight parameters to perform linear interpolation mixing on at least two sets of original test values, respectively, to obtain corresponding multiple sets of first mixed data; Inputting multiple sets of first mixed data into the environment simulation model, performing outlier screening and numerical optimization based on the attribute characteristics corresponding to the test environment, and obtaining multiple sets of second mixed data; The multiple sets of second mixed data are combined with the original test values ​​to obtain optimized short-time corona resistance life test values.

4. The method for accelerating prediction of corona resistance life of electromagnetic wire insulation layer according to claim 3, characterized in that: The generating of multiple sets of random weight parameters comprises: According to the real-time environment parameters of the test environment, selecting the data distribution type that matches the test sample in multiple test environments from the original data distribution library; Generate candidate random weight parameters matching respective distribution characteristics based on multiple data distribution types; Simulating the variation trend of the candidate random weight parameters according to the film material properties corresponding to the test sample; A rationality judgment is made based on the simulation results to obtain the random weight parameters of the final output; wherein the random weight parameters include at least: distribution shape parameters, weighted ratio parameters, random strategy selection parameters, and mixed ratio control parameters.

5. The method for accelerating prediction of corona resistance life of electromagnetic wire insulation layer according to claim 3, characterized in that: Before fusing the plurality of sets of second mixed data with the original test values ​​to obtain the optimized short-time corona resistance life test values, the method further includes: Determine a plurality of original noise types, and combine the plurality of original noise types into a plurality of groups of mixed noise types; Determine the noise intensity, fusion weight and injection data density corresponding to the test sample based on the type of thin film material corresponding to the test sample and the real-time operating parameters under the test environment; Based on the noise intensity, fusion weight and injected data density corresponding to the test sample, multiple groups of noise data corresponding to the respective mixed noise types are injected into the multiple groups of second mixed data and the original test values ​​to obtain multiple groups of third mixed data.

6. The method for accelerating prediction of corona resistance life of electromagnetic wire insulation layer according to claim 1, characterized in that: The method of constructing an accelerated prediction model based on the short-time corona resistance life test value comprises: Using the short-time corona resistance life test value and the environmental parameters corresponding to the test environment, constructing an accelerated prediction model training set; The life prediction formula is obtained based on the short-time corona resistance life test value fitting; the life prediction formula is expressed as: ; Wherein L is the short-time corona resistance life test value, is the electric field strength under the nth test environment, M is a constant, and T is the temperature; Using the life prediction formula as a constraint generation condition for the physical loss layer, generating corresponding dynamic constraint conditions; Based on the dynamic constraint condition, an initial prediction model is trained in the accelerated prediction model training set to obtain the accelerated prediction model; The accelerated prediction model is used to predict the corona resistance life value at a temperature and electric field strength less than the electrical aging threshold.

7. The method for accelerating prediction of corona resistance life of electromagnetic wire insulation layer according to claim 6, characterized in that: The generation of corresponding dynamic constraint conditions using the life prediction formula as the constraint generation condition of the physical loss layer includes: The corresponding relationship between the conductive current density and the applied electric field strength and the life prediction formula are mixed to generate the corresponding dynamic constraint condition; the dynamic constraint condition is expressed as the following formula: ； in, is the value of the physical loss function, is the predicted value of corona resistance life, is the i-th conductance current density, is the ith applied electric field strength in the nth test environment, A is a constant, is the total number of test environments.

8. The method for accelerating prediction of corona resistance life of electromagnetic wire insulation layer according to claim 1, characterized in that: The electric field strength used for testing in the test environment is less than the electrical aging threshold, and the electric field strength used for testing in the test environment is greater than the actual applied electric field strength in the actual environment.

9. An accelerated prediction system for corona resistance life of electromagnetic wire insulation layer, characterized in that: Applied to the scenario of corona resistance life prediction of thin film materials, the system includes: A collection unit is configured to collect the electrical conduction current density and applied electric field strength of a test sample in the electromagnetic wire insulation layer at different temperatures; the test sample corresponds to at least one thin film material to be tested; a determination unit configured to determine an electrical aging threshold corresponding to the test sample based on the electrical conduction current density and the applied electric field strength at different temperatures; A testing unit is configured to perform an accelerated corona resistance test based on the electrical aging threshold to obtain a short-time corona resistance life test value of the test sample under multiple test environments; wherein the multiple test environments respectively use different temperatures and different applied electric field strengths; and the applied electric field strengths used in the multiple test environments are all less than the target applied electric field strength corresponding to the electrical aging threshold; A construction unit configured to construct an accelerated prediction model based on the short-time corona resistance life test value; The prediction unit is configured to input the operating environment parameters of the target sample into the accelerated prediction model to obtain the target corona resistance life corresponding to the target sample; wherein the target sample corresponds to at least one thin film material to be tested; and the operating environment parameters at least include the real-time temperature and the real-time applied electric field strength in the operating environment of the target sample.

10. An accelerated prediction device for corona resistance life of electromagnetic wire insulation layer, characterized in that: The device at least comprises a corona resistance life prediction device, and the corona resistance life prediction device comprises at least one processor, a memory and an input-output unit; The memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the accelerated prediction method for the corona resistance life of the electromagnetic wire insulation layer according to any one of claims 1 to 8.

Images