Method and device for establishing residual life prediction model of insulated gate module

Abstract

The invention provides a method and device for establishing a residual life prediction model of an insulated gate module, and the method comprises the steps: collecting the power circulation data of the insulated gate module, calculating a health index based on the power circulation data, carrying out the preprocessing of an energy efficiency index corresponding to the power circulation data, and obtaining a prediction result. Forming a training set by the preprocessed energy efficiency indexes and health indexes; performing data enhancement on each index in the training set through discrete cosine transform to obtain a low-frequency feature, and fusing the low-frequency feature with features except the energy efficiency index in the power cycle data to obtain a composite feature; training a deep learning model including a self-attention mechanism by using the composite features, and adjusting model parameters in the deep learning model to obtain a first prediction model; and adjusting hyper-parameters in the first prediction model by using a swarm intelligence algorithm to obtain a second prediction model, and taking the second prediction model as a residual life prediction model. The low-frequency features are extracted, hyper-parameters are searched by using the swarm intelligence algorithm, and the prediction performance of the model is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of device life prediction, and particularly to a method and device for establishing a remaining life prediction model of an insulated gate module. Background Art

[0002] With the rapid development of the new energy vehicle market in China, the cost of insulated gate module IGBT power devices is almost second only to that of batteries, accounting for about 10% of the total vehicle cost. Therefore, it is particularly crucial to accurately evaluate the remaining service life of power devices, especially during the secondary utilization of power devices.

[0003] In related technologies, common IGBT life prediction technologies are mainly divided into two categories: physical model analysis methods and machine learning technologies. Currently, the following defects exist in these two technologies: The physical model method starts from multiple physical dimensions such as product structure, material properties, and environmental stress, and deeply explores the mechanism of fault formation at the end of life, and is more used for only fault analysis without remaining life prediction; The implementation of machine learning technology mines information by analyzing the historical data of devices, and the limitations of measurement devices lead to a decrease in the final prediction accuracy. The prior art CN116205140A realizes offline prediction but does not consider the impact of high-frequency data noise on the prediction results; And as the model becomes more complex, the number of hyperparameters also gradually increases. In the prior art CN1118746740A, hyperparameter tuning usually relies on manual experience, and the efficiency of manually adjusting parameters is low.

[0004] Based on the analysis of the development status of this technical field above, there is a lack of a solution in the existing technologies that performs data processing based on discrete cosine transform and automatically optimizes model hyperparameters using an intelligent swarm algorithm. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and device for establishing a remaining life prediction model of an insulated gate module, aiming to solve the above problems in the prior art.

[0006] According to the first aspect of the embodiments of the present invention, a method for establishing a remaining life prediction model of an insulated gate module is provided, including:

[0007] Collect power cycle data of the insulated gate module, calculate a health index based on the power cycle data, preprocess the energy efficiency index corresponding to the power cycle data, and form a training set with the preprocessed energy efficiency index and the health index;

[0008] Perform data augmentation on each index in the training set through discrete cosine transform to obtain low-frequency features, and fuse the low-frequency features with the features other than the energy efficiency index in the power cycle data to obtain composite features;

[0009] Train a deep learning model including a self-attention mechanism using composite features, and adjust the model parameters in the deep learning model to obtain a first prediction model, where the self-attention mechanism is used to capture temporal relationships;

[0010] Use a swarm intelligence algorithm to adjust the hyperparameters in the first prediction model to obtain a second prediction model, and use the second prediction model as the remaining useful life prediction model.

[0011] According to the second aspect of the embodiments of the present invention, there is provided an apparatus for establishing a remaining useful life prediction model of an insulated gate module, including:

[0012] A preprocessing module, configured to collect power cycle data of the insulated gate module, calculate a health index based on the power cycle data, preprocess the energy efficiency index corresponding to the power cycle data, and form a training set by combining the preprocessed energy efficiency index and the health index;

[0013] A feature enhancement module, configured to perform data enhancement on each index in the training set through discrete cosine transform to obtain low-frequency features, and fuse the low-frequency features with the features other than the energy efficiency index in the power cycle data to obtain composite features;

[0014] A model parameter adjustment module, configured to train a deep learning model including a self-attention mechanism using composite features, and adjust the model parameters in the deep learning model to obtain a first prediction model, where the self-attention mechanism is used to capture temporal relationships;

[0015] A hyperparameter adjustment module, configured to use a swarm intelligence algorithm to adjust the hyperparameters in the first prediction model to obtain a second prediction model, and use the second prediction model as the remaining useful life prediction model.

[0016] According to the third aspect of the embodiments of the present invention, there is provided an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where when the computer program is executed by the processor, the steps of the method for establishing a remaining useful life prediction model of an insulated gate module provided in the first aspect of the present disclosure are implemented.

[0017] According to the fourth aspect of the embodiments of the present invention, there is provided a computer-readable storage medium, on which an implementation program for information transmission is stored, and when the program is executed by a processor, the steps of the method for establishing a remaining useful life prediction model of an insulated gate module provided in the first aspect of the present disclosure are implemented.

[0018] The technical solutions provided by the embodiments of the present invention have the following beneficial effects: Data enhancement through discrete cosine transform helps to extract low-frequency features and suppress high-frequency noise that reflects short-term fluctuations and interference, so as to extract the main data information related to life prediction; Using a swarm intelligence algorithm to further train and automatically search for the optimal hyperparameter configuration can improve the prediction performance of the model.

[0019] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and do not limit the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in one or more embodiments of the present specification or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present specification. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0021] Figure 1 is a flowchart of a method for establishing a remaining life prediction model of an insulated gate module according to an embodiment of the present invention;

[0022] Figure 2 is a schematic diagram of an energy efficiency index according to an embodiment of the present invention;

[0023] Figure 3 is a schematic diagram of a self-attention mechanism deep learning model according to an embodiment of the present invention;

[0024] Figure 4 is a schematic diagram of a life prediction result according to an embodiment of the present invention;

[0025] Figure 5 is a schematic diagram of a device for establishing a remaining life prediction model of an insulated gate module according to an embodiment of the present invention;

[0026] Figure 6 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of the present specification, the following will clearly and completely describe the technical solutions in one or more embodiments of the present specification with reference to the drawings in one or more embodiments of the present specification. Obviously, the described embodiments are only a part of the embodiments of the present specification, rather than all the embodiments. Based on one or more embodiments of the present specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this document.

[0028] Method Embodiment

[0029] According to an embodiment of the present invention, a method for establishing a remaining life prediction model of an insulated gate module is provided. Figure 1 is a flowchart of a method for establishing a remaining life prediction model of an insulated gate module according to an embodiment of the present invention, as Figure 1As shown, the method for establishing the remaining life prediction model of the insulated gate module according to the embodiment of the present invention specifically includes:

[0030] In step S110, power cycle data of the insulated gate module is collected, a health index is calculated based on the power cycle data, the energy efficiency index corresponding to the power cycle data is preprocessed, and the preprocessed energy efficiency index and the health index are combined to form a training set, specifically including:

[0031] Use formula 1 to calculate the health index HI(t) of the corresponding device of the insulated gate module:

[0032]

[0033] Among them, Nf(t) represents the number of power cycle periods at time t, and Nf max represents the total number of power cycle periods.

[0034] Based on the power cycle data, use formula 2 to calculate the thermal resistance R th (t) of the corresponding device of the insulated gate module:

[0035]

[0036] Among them, T j (t) represents the junction temperature, T c (t) represents the case temperature, V ce (t) represents the collector-emitter saturation voltage, and I(t) represents the collector-emitter current;

[0037] Take the thermal resistance and the saturation voltage drop in the power cycle data as the energy efficiency index, Figure 2 is a schematic diagram of the energy efficiency index of the embodiment of the present invention. As Figure 2 shown, it shows the data collection schematic. Standardize the energy efficiency index to obtain the preprocessed energy efficiency index. Use formula 4 to represent the standardization process of the thermal resistance, and use the public

[0038] formula 5 to represent the standardization process of the saturation voltage drop:

[0039]

[0040]

[0041] Among them, R th * (t) represents the standardized thermal resistance, R thmin represents the minimum thermal resistance value measured in the entire power cycle experiment, R thmax represents the maximum thermal resistance value measured in the entire power cycle experiment, V ce (t) represents the unstandardized saturation voltage drop at time t, V ce *(t) represents the normalized saturation voltage drop, V cemin represents the minimum saturation voltage drop value measured during the entire power cycle experiment, V cemax represents the maximum saturation voltage drop value measured during the entire power cycle experiment.

[0042] Preferably, the corresponding validation set is generated in the same way to facilitate the subsequent verification of the model. This part of the content is carried out in a conventional manner within the field.

[0043] In step S120, discrete cosine transform is used to perform data augmentation on each index in the training set to obtain low-frequency features, and the low-frequency features are fused with the features in the power cycle data except for the energy efficiency index to obtain composite features, specifically including:

[0044] Use formula 3 to perform discrete cosine transform DCT on each index in the training set to obtain low-frequency features:

[0045]

[0046] where, X k represents the low-frequency features extracted after the enhancement transformation, x n represents the nth sample value in the training set, N represents the data length, and k represents the index in the frequency domain;

[0047] The low-frequency feature part usually contains the main information of the signal, can reflect the health change trend of the insulated gate module device during long-term use, and reduce the influence of high-frequency noise on model training;

[0048] Preferably, the low-frequency features can be further normalized and feature weighted. Use formula 6 to normalize the low-frequency features:

[0049]

[0050] where, X k * represents the normalized low-frequency features, X min represents the minimum value in the low-frequency feature sequence, X max represents the maximum value in the low-frequency feature sequence. After normalization, the data has a unified dimension, which is convenient for subsequent model processing;

[0051] Feature weighting can emphasize the influence of important frequency domain features in the low-frequency features, and is adjusted using adaptive weights to emphasize the frequency features related to the device health state, thereby improving the contribution to the remaining life prediction. In the embodiment of the present invention, it is implemented through a weight generation network based on the attention mechanism. This network takes the low-frequency features as input, outputs the weights of each frequency component, and uses the channel attention mechanism to capture the correlation between different frequency classifications;

[0052] In an embodiment of the present invention, low-frequency features are fused with indicators such as the collector-emitter voltage in power cycle data to form composite features in a splicing manner.

[0053] In step S130, a deep learning model including a self-attention mechanism is trained using the composite features, and the model parameters in the deep learning model are adjusted to obtain a first prediction model. The self-attention mechanism is used to capture temporal relationships and specifically includes:

[0054] The composite features are input into a Transformer model including a self-attention mechanism for training. Figure 3 is a schematic diagram of the self-attention mechanism deep learning model according to an embodiment of the present invention. As Figure 3 shown, it shows the model encoder part and decoder part. By weighting the composite features at different time steps through the self-attention mechanism, features with different degrees of importance are obtained and an initial prediction result is output.

[0055] The model parameters of the Transformer model are adjusted according to the initial prediction result in a commonly used manner in the field. The model parameters refer to variables or coefficients learned inside the model during the general training process, which are automatically adjusted through an optimization algorithm during the training process, such as bias terms, connection weights, and weight matrices related to the attention mechanism, etc., to obtain a first prediction model.

[0056] In step S140, a swarm intelligence algorithm is used to adjust the hyperparameters in the first prediction model to obtain a second prediction model, and the second prediction model is used as the remaining life prediction model. Specifically, it includes:

[0057] The particle swarm optimization algorithm is used as the swarm intelligence algorithm to adjust the hyperparameters. The hyperparameters include the hidden layer size, learning rate, and number of attention heads.

[0058] The method further includes:

[0059] In step S150, the remaining life prediction model is used to predict the remaining life duration of the insulated gate module. Specifically, it includes:

[0060] According to the prediction result, the remaining working cycle or usage time of the device is provided to guide the subsequent maintenance and replacement of the device. Preferably, by collecting the user's usage habits online through sensors, the user's usage habits are input into the trained lightweight prediction model, and a preset time is set to regularly and real-time output the adjustment difference of the remaining life of the insulated gate module. If the adjustment difference output is -20h and the originally predicted remaining life duration is 530h, then the final remaining life duration is adjusted to 510h.

[0061] Figure 4 is a schematic diagram of the life prediction result according to an embodiment of the present invention. As Figure 4As shown, the prediction results within a time step interval are presented, and the error between the predicted value and the true value is small.

[0062] In summary, for the problems existing in the current situation, the method for establishing the remaining life prediction model of the insulated gate module in this invention helps to extract low-frequency features through discrete cosine transform and suppress high-frequency noise that reflects short-term fluctuations and interference, so as to extract the main data information related to life prediction; the normalization process in the data preprocessing ensures the consistency of the data; the low-frequency features are fused with the features in the power cycle data except the energy efficiency index to obtain composite features, so as to provide richer device state information and provide high-quality data input for the model; the Transformer model including the self-attention mechanism is selected as the deep learning model to weight the attention to important features enhanced by different time steps, capture the complex temporal relationships in the data to improve the prediction accuracy; the swarm intelligence algorithm is used to further train and automatically search for the optimal hyperparameter configuration to improve the prediction performance of the model.

[0063] Device Embodiment

[0064] According to an embodiment of the present invention, there is provided an apparatus for establishing a remaining life prediction model of an insulated gate module. Figure 5 It is a schematic diagram of the apparatus for establishing the remaining life prediction model of the insulated gate module according to an embodiment of the present invention, as Figure 5 shown. The apparatus for establishing the remaining life prediction model of the insulated gate module according to an embodiment of the present invention specifically includes:

[0065] A preprocessing module 50, configured to collect power cycle data of the insulated gate module, calculate a health index based on the power cycle data, preprocess the energy efficiency index corresponding to the power cycle data, and form a training set with the preprocessed energy efficiency index and the health index. Specifically, it is used for:

[0066] Use Equation 1 to calculate the health index HI(t) of the device corresponding to the insulated gate module:

[0067]

[0068] where Nf(t) represents the number of power cycle periods at time t, and Nf max represents the total number of power cycle periods.

[0069] Based on the power cycle data, use Equation 2 to calculate the thermal resistance R th (t) of the device corresponding to the insulated gate module:

[0070]

[0071] where T j (t) represents the junction temperature, and T c(t) represents the case temperature, V ce (t) represents the collector-emitter saturation voltage, and I(t) represents the collector-emitter current;

[0072] Taking the saturation voltage drop in the thermal resistance and power cycle data as the energy efficiency index, standardizing the energy efficiency index to obtain the preprocessed energy efficiency index.

[0073] The feature enhancement module 52 is used to perform data enhancement on each index in the training set through discrete cosine transform to obtain low-frequency features, and fuse the low-frequency features with the features in the power cycle data except the energy efficiency index to obtain composite features. Specifically, it is used for:

[0074] Performing discrete cosine transform on each index in the training set using formula 3 to obtain low-frequency features:

[0075]

[0076] where X k represents the low-frequency features extracted after enhanced transformation, x n represents the nth sample value in the training set, N represents the data length, and k represents the index in the frequency domain.

[0077] The model parameter adjustment module 54 is used to train a deep learning model including a self-attention mechanism using the composite features, and adjust the model parameters in the deep learning model to obtain the first prediction model. Among them, the self-attention mechanism is used to capture the temporal relationship. Specifically, it is used for:

[0078] Inputting the composite features into a Transformer model including a self-attention mechanism for training, weighting the composite features at different time steps through the self-attention mechanism to obtain importance-distinguished features and output the initial prediction result;

[0079] Adjusting the model parameters of the Transformer model according to the initial prediction result to obtain the first prediction model.

[0080] The hyperparameter adjustment module 56 is used to adjust the hyperparameters in the first prediction model using a swarm intelligence algorithm to obtain the second prediction model, and use the second prediction model as the remaining life prediction model. Specifically, it is used for:

[0081] Using the particle swarm optimization algorithm as the swarm intelligence algorithm to adjust the hyperparameters, where the hyperparameters include the hidden layer size, learning rate, and number of attention heads.

[0082] The method further includes:

[0083] An application module, which is used to predict the remaining life duration of the insulated gate module using the remaining life prediction model.

[0084] In summary, in view of the existing problems in the current situation, the device for establishing the remaining life prediction model of the insulated gate module of the present invention uses discrete cosine transform for data augmentation, which helps to extract low-frequency features and suppress high-frequency noise that reflects short-term fluctuations and interference, so as to extract the data information mainly related to life prediction; the normalization process in the data preprocessing ensures the consistency of the data; the low-frequency features are fused with the features in the power cycle data except the energy efficiency index to obtain composite features, so as to provide richer device state information and provide high-quality data input for the model; the Transformer model including the self-attention mechanism is selected as the deep learning model to weight the importance of data augmentation at different time steps for features important for prediction, and capture the complex temporal relationships in the data to improve the prediction accuracy; the swarm intelligence algorithm is used to further train and automatically search for the optimal hyperparameter configuration to improve the prediction performance of the model.

[0085] Embodiment of Electronic Device

[0086] Figure 6 FIG. is a schematic diagram of an electronic device according to an embodiment of the present invention. The electronic device 600 may include at least one processor 610 and a memory 620. The processor 610 may execute instructions stored in the memory 620. The processor 610 is communicatively connected to the memory 620 via a data bus. In addition to the memory 620, the processor 610 may also be communicatively connected to an input device 630, an output device 640, and a communication device 650 via the data bus.

[0087] The processor 610 may be any conventional processor, such as a commercially available CPU. The processor may also include, for example, a Graphic Process Unit (GPU), a Field Programmable Gate Array (FPGA), a System on Chip (SOC), an Application Specific Integrated Circuit (ASIC), or a combination thereof.

[0088] The memory 620 may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disk.

[0089] In an embodiment of the present disclosure, executable instructions are stored in the memory 620, and the processor 610 can read the executable instructions from the memory 620 and execute the instructions to implement all or part of the steps of the method for establishing the remaining life prediction model of the insulated gate module in any of the above exemplary embodiments.

[0090] Embodiment of computer-readable storage medium

[0091] In addition to the above methods and devices, an exemplary embodiment of the present disclosure may also be a computer program product or a computer-readable storage medium storing the computer program product. The computer product includes computer program instructions that can be executed by a processor to implement all or part of the steps described in the method for establishing the remaining life prediction model of the insulated gate module in any of the above exemplary embodiments.

[0092] The computer program product can be written in any combination of one or more programming languages for executing the program code of the operations of the embodiments of the present application. The programming languages include object-oriented programming languages such as Java, C++, etc., and also include conventional procedural programming languages such as the "C" language or similar programming languages and scripting languages (such as Python). The program code can be executed entirely on the user computing device, partially on the user device, executed as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0093] The computer-readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the readable storage medium include: a static random access memory (SRAM) with one or more wire electrical connections, an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk, or any suitable combination of the above.

[0094] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for establishing a residual life prediction model for an insulating gate module, characterized in that: include: Collecting power cycle data of the insulation gate module, calculating a health index based on the power cycle data, preprocessing the energy efficiency index corresponding to the power cycle data, and forming a training set with the preprocessed energy efficiency index and the health index; Performing data enhancement on each indicator in the training set by discrete cosine transform to obtain low-frequency features, and fusing the low-frequency features with features in the power cycle data except the energy efficiency indicator to obtain composite features; Using the composite feature to train a deep learning model including a self-attention mechanism, and adjusting model parameters in the deep learning model to obtain a first prediction model, wherein the self-attention mechanism is used to capture a temporal relationship; A swarm intelligence algorithm is used to adjust the hyperparameters in the first prediction model to obtain a second prediction model, and the second prediction model is used as the remaining life prediction model.

2. The method according to claim 1, characterized in that The method further comprises: The remaining life prediction model is used to predict the remaining life of the insulation gate module.

3. The method according to claim 1, characterized in that The calculating of the health index based on the power cycle data specifically includes: Use formula 1 to calculate the health index HI(t) of the device corresponding to the insulated gate module: Where Nf(t) represents the number of power cycles at time t, Nf max Indicates the total number of power cycle periods.

4. The method according to claim 1, characterized in that: The preprocessing of the energy efficiency index corresponding to the power cycle data specifically includes: Based on the power cycle data, the thermal resistance R of the corresponding device of the insulated gate module is calculated using Formula 2. th (t): Among them, T j (t) represents the junction temperature, T c (t) represents the shell temperature, V ce (t) represents the collector-emitter saturation voltage, I(t) represents the collector-emitter current; The thermal resistance and the saturated voltage drop in the power cycle data are used as energy efficiency indicators, and the energy efficiency indicators are standardized to obtain pre-processed energy efficiency indicators.

5. The method according to claim 1, characterized in that The method of performing data enhancement on each indicator in the training set by discrete cosine transform to obtain low-frequency features specifically includes: Use formula 3 to perform discrete cosine transform on each indicator in the training set to obtain low-frequency features: Among them, X k represents the low-frequency features extracted after enhanced transformation, x n It represents the nth sample value in the training set, N represents the data length, and k represents the index in the frequency domain.

6. The method according to claim 1, characterized in that The using the composite feature to train a deep learning model including a self-attention mechanism, and adjusting model parameters in the deep learning model to obtain a first prediction model specifically includes: Inputting the composite features into a Transformer model including a self-attention mechanism for training, weighting composite features of different time steps through the self-attention mechanism, obtaining importance distinguishing features and outputting initial prediction results; The model parameters of the Transformer model are adjusted according to the initial prediction result to obtain a first prediction model.

7. The method according to claim 1, characterized in that The using of a swarm intelligence algorithm to adjust the hyperparameters in the first prediction model specifically includes: using a particle swarm optimization algorithm as a swarm intelligence algorithm to adjust the hyperparameters, wherein the hyperparameters include a hidden layer size, a learning rate, and a number of attention heads.

8. A device for establishing a residual life prediction model of an insulating gate module, characterized in that: include: A preprocessing module, used for collecting power cycle data of the insulation gate module, calculating the health index based on the power cycle data, preprocessing the energy efficiency index corresponding to the power cycle data, and forming a training set with the preprocessed energy efficiency index and the health index; A feature enhancement module, used to perform data enhancement on each indicator in the training set by discrete cosine transform to obtain low-frequency features, and fuse the low-frequency features with features in the power cycle data except the energy efficiency indicator to obtain composite features; A model parameter adjustment module, used to use the composite feature to train a deep learning model including a self-attention mechanism, and adjust model parameters in the deep learning model to obtain a first prediction model, wherein the self-attention mechanism is used to capture a temporal relationship; A hyperparameter adjustment module is used to use a swarm intelligence algorithm to adjust the hyperparameters in the first prediction model to obtain a second prediction model, and use the second prediction model as the remaining life prediction model.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the method for establishing a remaining life prediction model of an insulating gate module as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an implementation program for information transmission, and when the program is executed by a processor, the steps of the method for establishing the remaining life prediction model of the insulation gate module according to any one of claims 1 to 7 are implemented.

Images