High-voltage power module chip temperature inversion method, system, device and medium

Abstract

The invention discloses a high-voltage power module chip temperature inversion method, system and device and a medium, and the method comprises the steps: constructing a high-fidelity finite element model of a high-voltage power module, carrying out simulation under the boundary conditions of different current levels, different radiator temperatures and the like, obtaining a chip temperature and a housing temperature, and constructing a data set based on the obtained simulation data. A multi-input multi-output deep neural network is established, and a non-linear mapping relation between measurable point temperature and chip temperature is obtained through training of a simulation data set. An experiment is carried out to obtain a small sample experiment data set of the chip temperature and the shell temperature under the actual working condition, the small sample experiment data set is utilized to carry out fine adjustment on the trained deep neural network, the nonlinear relation is corrected, and the chip temperature inversion accuracy is improved. Therefore, the problem that the chip temperature of the high-voltage power module is difficult to accurately and efficiently obtain in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the technical field of high-voltage power module state analysis, and in particular to a high-voltage power module chip temperature inversion method, system, equipment and medium. Background Art

[0002] High-voltage power modules occupy an important position in modern power electronics applications and are widely used in many fields such as DC transmission, railway traction, and electric vehicles. With the continuous advancement of technology and the increasing demand for higher power density and integrated converters, the application of multi-chip power modules is becoming more and more popular. However, these modules face the problem of high-temperature failure during operation, which is one of the main causes of their failure. The chip is located inside the high-voltage power module and is encapsulated by insulating materials. Its temperature is difficult to accurately obtain by conventional means. Quickly and accurately inverting the chip temperature of the power module is crucial to achieve effective thermal management, life estimation, and condition monitoring.

[0003] Traditional methods for obtaining module chip temperature mainly include direct measurement method, thermal circuit calculation method, etc. The direct measurement method is easy to damage the module itself and is difficult to promote widely; the thermal circuit calculation method is complicated, the thermal circuit model is difficult to obtain, and the calculation results often deviate greatly from the actual results, making it difficult to widely realize engineering applications. Summary of the invention

[0004] The present application provides a high-voltage power module chip temperature inversion method, system, device and medium, which are used to solve the problem that it is difficult to accurately and efficiently obtain the chip temperature of the high-voltage power module in the prior art.

[0005] In view of this, the first aspect of the present application provides a high-voltage power module chip temperature inversion method, the method comprising:

[0006] Constructing a high-fidelity finite element model of the high-voltage power module to be inverted, and verifying the correctness of the high-fidelity finite element model;

[0007] Based on the high-fidelity finite element model after correctness verification, simulation is performed according to preset boundary conditions to obtain chip temperature and housing temperature for constructing a data set, wherein the preset boundary conditions include: different current levels and different heat sink temperatures;

[0008] Building a multi-input multi-output deep neural network, and training the multi-input multi-output deep neural network through the preprocessed data set to obtain a deep neural network for characterizing the nonlinear mapping relationship between the measurable point temperature and the chip temperature;

[0009] After building a two-level inverter using a high-voltage power module and setting the control algorithm and operating parameters, the load current is controlled by adjusting the load resistance, and several sets of chip temperature and shell temperature sample experimental data sets are obtained by detection at the same time;

[0010] Based on the transfer learning theory, the sample experimental data set is used to fine-tune the parameters of the trained deep neural network, thereby correcting the nonlinear relationship and obtaining the deep neural network that is ultimately used to invert the chip temperature of the high-voltage power module.

[0011] Optionally, constructing a high-fidelity finite element model of the high-voltage power module to be inverted specifically includes:

[0012] Obtaining the geometric dimensions of the high-voltage power module to be inverted and the material characteristic parameters of each component of the high-voltage power module to be inverted, wherein the geometric dimensions include: chip size, heat sink structure, welding layer thickness, and packaging material thickness;

[0013] According to the geometric dimensions and the material characteristic parameters, a high-fidelity finite element model of the high-voltage power module to be inverted is constructed using finite element analysis software.

[0014] Optionally, the verifying the correctness of the high-fidelity finite element model specifically includes:

[0015] Using network units of different sizes to segment the chip, ceramic and copper substrate in the high-fidelity finite element model respectively;

[0016] The average temperature and maximum temperature of the chip surface obtained by infrared camera testing and the finite element simulation calculation results are used to verify the correctness of the high-fidelity finite element model.

[0017] Optionally, the building of a multi-input and multi-output deep neural network specifically includes:

[0018] A multi-layer perceptron structure is used as the network structure to build a multi-input and multi-output deep neural network, where the network input includes: several measurable point temperatures, and the network output includes several chip temperatures.

[0019] A second aspect of the present application provides a high-voltage power module chip temperature inversion system, the system comprising:

[0020] A construction unit, used to construct a high-fidelity finite element model of the high-voltage power module to be inverted, and verify the correctness of the high-fidelity finite element model;

[0021] A simulation unit, used to simulate based on the high-fidelity finite element model after correctness verification and according to preset boundary conditions, to obtain chip temperature and housing temperature for constructing a data set, wherein the preset boundary conditions include: different current levels and different heat sink temperatures;

[0022] A training unit, used for building a multi-input multi-output deep neural network, and training the multi-input multi-output deep neural network through the preprocessed data set to obtain a deep neural network for characterizing the nonlinear mapping relationship between the measurable point temperature and the chip temperature;

[0023] The acquisition unit is used to build a two-level inverter using a high-voltage power module and set the control algorithm and working parameters, and then control the load current by adjusting the load resistance, and at the same time, detect and obtain several groups of sample experimental data sets of chip temperature and shell temperature;

[0024] The adjustment unit is used to fine-tune the parameters of the trained deep neural network based on the transfer learning theory and the sample experimental data set, so as to correct the nonlinear relationship and obtain the deep neural network finally used for inverting the chip temperature of the high-voltage power module.

[0025] Optionally, constructing a high-fidelity finite element model of the high-voltage power module to be inverted specifically includes:

[0026] Obtaining the geometric dimensions of the high-voltage power module to be inverted and the material characteristic parameters of each component of the high-voltage power module to be inverted, wherein the geometric dimensions include: chip size, heat sink structure, welding layer thickness, and packaging material thickness;

[0027] According to the geometric dimensions and the material characteristic parameters, a high-fidelity finite element model of the high-voltage power module to be inverted is constructed using finite element analysis software.

[0028] Optionally, the verifying the correctness of the high-fidelity finite element model specifically includes:

[0029] Using network units of different sizes to segment the chip, ceramic and copper substrate in the high-fidelity finite element model respectively;

[0030] The average temperature and maximum temperature of the chip surface obtained by infrared camera testing and the finite element simulation calculation results are used to verify the correctness of the high-fidelity finite element model.

[0031] Optionally, the building of a multi-input and multi-output deep neural network specifically includes:

[0032] A multi-layer perceptron structure is used as the network structure to build a multi-input and multi-output deep neural network, where the network input includes: several measurable point temperatures, and the network output includes several chip temperatures.

[0033] A third aspect of the present application provides a high-voltage power module chip temperature inversion method device, the device comprising a processor and a memory:

[0034] The memory is used to store program code and transmit the program code to the processor;

[0035] The processor is used to execute the steps of the high-voltage power module chip temperature inversion method as described in the first aspect according to the instructions in the program code.

[0036] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute the high-voltage power module chip temperature inversion method described in the first aspect above.

[0037] It can be seen from the above technical solutions that this application has the following advantages:

[0038] The present application provides a method for inverting the chip temperature of a high-voltage power module. First, a high-fidelity finite element model of the high-voltage power module is constructed. Simulations are carried out under boundary conditions such as different current levels and different radiator temperatures to obtain the chip temperature and the housing temperature, and a data set is constructed based on the obtained simulation data. Then, a multi-input and multi-output deep neural network is built, and the simulation data set is used to train the nonlinear mapping relationship between the measurable point temperature and the chip temperature. Finally, experiments are carried out to obtain a small sample experimental data set of the chip temperature and the housing temperature under actual working conditions. The small sample experimental data set is used to fine-tune the trained deep neural network, correct the nonlinear relationship, and improve the accuracy of chip temperature inversion.

[0039] Compared with the prior art, the present application constructs a nonlinear relationship between the measurable point temperature and the chip temperature of the high-voltage power module based on a neural network, and uses the measurable point temperature to efficiently and accurately invert the chip temperature, which is easier to promote on a large scale than the traditional method. This solves the problem that the prior art is difficult to accurately and efficiently obtain the chip temperature of the high-voltage power module. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic flow chart of a high-voltage power module chip temperature inversion method provided in an embodiment of the present application;

[0041] Figure 2 A finite element model of a high-voltage power module provided in an embodiment of the present application;

[0042] Figure 3The mesh division result of the welding type IGBT provided in the embodiment of the present application;

[0043] Figure 4 This is the high-fidelity model verification result provided in the embodiments of this application;

[0044] Figure 5 The transient response curve of the high-voltage power module chip temperature provided in the embodiment of the present application;

[0045] Figure 6 It is the MLP neural network structure provided in the embodiments of this application;

[0046] Figure 7 It is the training process loss function provided in the embodiment of the present application;

[0047] Figure 8 A schematic diagram of a two-level inverter experimental platform provided in an embodiment of the present application;

[0048] Fig. 9 The load current waveform provided in the embodiment of the present application;

[0049] Fig.10 This is a parameter fine-tuning principle diagram provided in the embodiments of the present application;

[0050] Fig.11 It is the chip temperature inversion error provided in the embodiment of the present application;

[0051] Fig.12 This is a schematic diagram of the structure of a high-voltage power module chip temperature inversion system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to enable those skilled in the art to better understand the solution of the present application, 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 only 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.

[0053] See also Figure 1 A high-voltage power module chip temperature inversion method provided in an embodiment of the present application includes:

[0054] Step 101: construct a high-fidelity finite element model of the high-voltage power module to be inverted, and verify the correctness of the high-fidelity finite element model.

[0055] In one embodiment, a high-fidelity finite element model of the high-voltage power module to be inverted is constructed as follows:

[0056] The geometric dimensions of the high-voltage power module to be inverted and the material characteristic parameters of each component of the high-voltage power module to be inverted are obtained, wherein the geometric dimensions include: chip size, heat sink structure, welding layer thickness, and packaging material thickness; according to the geometric dimensions and material characteristic parameters, a high-fidelity finite element model of the high-voltage power module to be inverted is constructed using finite element analysis software.

[0057] It should be noted that when constructing a finite element model of a high-voltage power module, the geometric dimensions of the module must be accurately measured first. These dimensions include chip size, heat sink structure, solder layer thickness, packaging material thickness, etc. The accuracy of the geometric dimensions can be ensured by using high-precision measuring equipment, such as a coordinate measuring machine (CMM) or a laser scanner. In addition, in order to build a high-fidelity model, the material property parameters of each component of the high-voltage power module need to be obtained. Including the thermal conductivity, specific heat capacity, density, etc. of chip material silicon, the thermal conductivity and thermal expansion coefficient of packaging materials (such as ceramics and copper), and the thermal conductivity and shear modulus of welding materials. These parameters can be obtained through literature search, data provided by the manufacturer, or through experimental testing. After obtaining the geometric dimensions and material parameters, a high-fidelity finite element model is established using finite element analysis software (COMSOL Multiphysics). The geometry and material properties of each component must be considered in the model, and appropriate boundary conditions must be applied, including current level and heat sink temperature. This application takes a commercial high-voltage power module (FF150R12ME3G) as an example. The dimensions and materials of the module are obtained from the data sheet. The high-fidelity model is obtained by drawing with finite element analysis software. Figure 2 shown.

[0058] In one embodiment, the correctness of the high-fidelity finite element model is verified as follows:

[0059] The chip, ceramic and copper substrates in the high-fidelity finite element model are segmented using network units of different sizes. The average and maximum temperatures of the chip surface tested by an infrared camera and the finite element simulation results are used to verify the correctness of the high-fidelity finite element model.

[0060] It should be noted that the high-fidelity model of the high-voltage power module needs to be segmented before analysis. The finite element method needs to divide the model into units with simple geometric shapes. Reasonable mesh division is conducive to improving the accuracy and convergence of the results. Since the chip size is smaller than other components and has a greater impact on the calculation results, smaller mesh units are used for sweeping, and the maximum unit size is set to 0.5mm. The thermal conductivity of ceramic and copper substrates is large, and their temperature distribution is relatively uniform, so larger network units can be used for division. After segmentation, there are a total of 807,656 mesh units and 199,540 vertices. The segmentation results of the finite element simulation model of the high-voltage power module are as follows: Figure 3 shown.

[0061] In order to verify the accuracy of the high-fidelity model, the average and maximum temperatures of the chip surfaces obtained by the infrared camera test and the finite element simulation calculation results are listed in Table 1. The average surface temperature of the three parallel chips on the lower tube obtained by finite element calculation has an error of less than 0.5°C compared with the measured value; the maximum deviation between the calculated maximum surface temperature of the three parallel chips and the measured value is 1.7°C. The bonding wires on the chip surface affect the measured maximum surface temperature results. The surface temperature tested by the infrared camera is affected by the effect of the heat-reflective black paint spraying. The surface temperature tested by the infrared camera is as follows: Figure 4 Therefore, it is believed that under the heat sink and heat dissipation conditions used in this application, the high-fidelity model of the high-voltage power module is accurate and reasonable.

[0062] Table 1 Chip surface temperature comparison

[0063]

[0064] Step 102: Based on the high-fidelity finite element model after correctness verification, simulation is performed according to preset boundary conditions to obtain chip temperature and housing temperature for constructing a data set, wherein the preset boundary conditions include: different current levels and different heat sink temperatures.

[0065] It should be noted that since training a neural network and obtaining its nonlinear mapping function requires a large amount of data as support, we select the common operating current level and heat sink temperature of high-voltage power modules and construct a data set under this condition.

[0066] This application selects working current levels of 25A, 50A, 75A, 100A, 125A and 150A, and radiator temperatures of 0°C, 5°C, 10°C, 15°C, 20°C, 25°C and 30°C. A total of 36 groups of working condition simulation results are obtained by combining two working conditions. Since the transient temperature rise curve is simulated, each group of working condition simulation results can be resampled to obtain M groups. The M value can be determined according to the size of the transient temperature rise time. This application selects M as 8000, that is, 8000 groups of training data can be obtained for each group of working conditions. In this way, 36×8000=288000 groups of training data can be obtained in the end. The simulation results with a current of 50A and a radiator temperature of 25°C are selected, such as Figure 5 shown.

[0067] In order to find the nonlinear mapping relationship, the 12 substrate positions corresponding to the high-voltage power module chip are selected as measurable points, and a structure of 12 measurable points mapping 12 chip temperatures is constructed, that is, a 24-bit one-dimensional array. The first 12 bits represent the measurable point temperature, and the last 12 bits represent the chip temperature.

[0068] Step 103: construct a multi-input multi-output deep neural network, and train the multi-input multi-output deep neural network through the preprocessed data set to obtain a deep neural network for characterizing the nonlinear mapping relationship between the measurable point temperature and the chip temperature.

[0069] In one embodiment, a multi-input and multi-output deep neural network is constructed as follows:

[0070] A multi-layer perceptron structure is used as the network structure to build a multi-input and multi-output deep neural network, where the network input includes: several measurable point temperatures, and the network output includes several chip temperatures.

[0071] It should be noted that a multiple-input multiple-output (MIMO) deep neural network is designed. The network input includes 12 measurable point temperatures, and the network output is 12 chip temperatures. The network structure adopts a multi-layer perceptron (MLP) structure, such as Figure 6 shown.

[0072] Then, the simulation data is preprocessed, including normalization and denoising. Normalization can make the data on the same scale, which is convenient for neural network training; denoising can improve the quality of data and training effect. Use the preprocessed simulation data set to train the deep neural network. During the training process, the cross-validation method can be used to prevent overfitting. Select a suitable loss function (such as mean square error MSE) and optimization algorithm (such as Adam), adjust the network parameters, and ensure that the network converges to the optimal solution. The loss function value of the training process is as follows: Figure 7 As shown. After the training is completed, a deep neural network that can represent nonlinear mapping relationships is obtained.

[0073] Step 104: After building a two-level inverter using a high-voltage power module and setting a control algorithm and operating parameters, the load current is controlled by adjusting the load resistance, and at the same time, detection is performed to obtain several groups of sample experimental data sets of chip temperature and housing temperature.

[0074] It should be noted that the two-level inverter is built using the high-voltage power module model FF150R12ME3G. Figure 8 As shown. The inverter adopts a sinusoidal pulse width modulation control algorithm, the DC voltage Udc is 200V, the control switching frequency fsw is 4kHz, the carrier frequency f0 is 1Hz, and the load inductance L is 5mH. The high-voltage power module needs to remove the silicone test chip temperature. The module cannot work in a high-voltage state, so the DC voltage is maintained at 200V. The load current is controlled by adjusting the load resistance. The load resistance R selected in this application is 0.25Ω, and the modulated load current is as follows Fig. 9 As shown, the maximum current can reach about 80A.

[0075] 100 sets of experimental data were obtained through measurements with sensors and infrared thermometers to construct a small sample data set for the experiment.

[0076] Step 105: Based on the transfer learning theory, the sample experimental data set is used to fine-tune the parameters of the trained deep neural network, so as to correct the nonlinear relationship and obtain the deep neural network finally used to invert the chip temperature of the high-voltage power module.

[0077] It should be noted that transfer learning is a machine learning method that transfers the knowledge of a trained model to a new task. For this project, you can use the deep neural network model trained on simulation data and fine-tune it with a small amount of experimental data to improve the prediction accuracy of the model under actual working conditions. Load the deep neural network model trained on simulation data and freeze some parameters (the weights of the first few layers), and only fine-tune the last few layers, such as Fig.10 shown.

[0078] In this way, the basic structure and most of the knowledge of the model can be retained, while the model can be adaptively adjusted through small sample experimental data. Use small sample experimental data to fine-tune the deep neural network. Select appropriate learning rate and training strategy to ensure the convergence of the model on small sample data. Through the feedback of experimental data, gradually adjust the model parameters to improve the accuracy of chip temperature inversion. In this application, the chip inversion error can be reduced from 7.5% to 2.5%, such as Fig.11 As shown, the effect is significant.

[0079] The present application provides a method for inverting the chip temperature of a high-voltage power module. First, a high-fidelity finite element model of the high-voltage power module is constructed. Simulations are carried out under boundary conditions such as different current levels and different radiator temperatures to obtain the chip temperature and the shell temperature, and a data set is constructed based on the obtained simulation data. Then, a multi-input and multi-output deep neural network is built, and the simulation data set is used to train the nonlinear mapping relationship between the measurable point temperature and the chip temperature. Finally, experiments are carried out to obtain a small sample experimental data set of the chip temperature and the shell temperature under actual working conditions. The small sample experimental data set is used to fine-tune the trained deep neural network, correct the nonlinear relationship, and improve the accuracy of chip temperature inversion.

[0080] The above is a high-voltage power module chip temperature inversion method provided in an embodiment of the present application, and the following is a high-voltage power module chip temperature inversion system provided in an embodiment of the present application.

[0081] See also Fig.12 A high-voltage power module chip temperature inversion system provided in an embodiment of the present application includes:

[0082] The construction unit 201 is used to construct a high-fidelity finite element model of the high-voltage power module to be inverted, and verify the correctness of the high-fidelity finite element model.

[0083] The simulation unit 202 is used to simulate based on the high-fidelity finite element model after correctness verification and according to preset boundary conditions to obtain chip temperature and housing temperature for constructing a data set, wherein the preset boundary conditions include: different current levels and different heat sink temperatures.

[0084] The training unit 203 is used to build a multi-input multi-output deep neural network, and train the multi-input multi-output deep neural network through the preprocessed data set to obtain a deep neural network for characterizing the nonlinear mapping relationship between the measurable point temperature and the chip temperature.

[0085] The acquisition unit 204 is used to build a two-level inverter using a high-voltage power module and set a control algorithm and working parameters, control the load current by adjusting the load resistance, and simultaneously detect to obtain several groups of sample experimental data sets of chip temperature and housing temperature.

[0086] The adjustment unit 205 is used to fine-tune the parameters of the trained deep neural network based on the transfer learning theory and the sample experimental data set, so as to correct the nonlinear relationship and finally obtain the deep neural network used to invert the chip temperature of the high-voltage power module.

[0087] Furthermore, an embodiment of the present application also provides a high-voltage power module chip temperature inversion device, the device comprising a processor and a memory:

[0088] The memory is used to store program code and transmit the program code to the processor;

[0089] The processor is used to execute the steps of the high-voltage power module chip temperature inversion method as described in the above method embodiment according to the instructions in the program code.

[0090] Furthermore, a computer-readable storage medium is provided in an embodiment of the present application, and the computer-readable storage medium is used to store program code, and the program code is used to execute the high-voltage power module chip temperature inversion method described in the above method embodiment.

[0091] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0092] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein, for example. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0093] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0094] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0095] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0096] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0097] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program codes.

[0098] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A high-voltage power module chip temperature inversion method, characterized in that: include: Constructing a high-fidelity finite element model of the high-voltage power module to be inverted, and verifying the correctness of the high-fidelity finite element model; Based on the high-fidelity finite element model after correctness verification, simulation is performed according to preset boundary conditions to obtain chip temperature and housing temperature for constructing a data set, wherein the preset boundary conditions include: different current levels and different heat sink temperatures; Building a multi-input multi-output deep neural network, and training the multi-input multi-output deep neural network through the preprocessed data set to obtain a deep neural network for characterizing the nonlinear mapping relationship between the measurable point temperature and the chip temperature; After building a two-level inverter using a high-voltage power module and setting the control algorithm and operating parameters, the load current is controlled by adjusting the load resistance, and several sets of chip temperature and shell temperature sample experimental data sets are obtained by detection at the same time; Based on the transfer learning theory, the sample experimental data set is used to fine-tune the parameters of the trained deep neural network, thereby correcting the nonlinear relationship and obtaining the deep neural network that is ultimately used to invert the chip temperature of the high-voltage power module.

2. The high-voltage power module chip temperature inversion method according to claim 1, characterized in that: The construction of a high-fidelity finite element model of the high-voltage power module to be inverted specifically includes: Obtaining the geometric dimensions of the high-voltage power module to be inverted and the material characteristic parameters of each component of the high-voltage power module to be inverted, wherein the geometric dimensions include: chip size, heat sink structure, welding layer thickness, and packaging material thickness; According to the geometric dimensions and the material characteristic parameters, a high-fidelity finite element model of the high-voltage power module to be inverted is constructed using finite element analysis software.

3. The high-voltage power module chip temperature inversion method according to claim 1, characterized in that: The verification of the correctness of the high-fidelity finite element model specifically includes: Using network units of different sizes to segment the chip, ceramic and copper substrate in the high-fidelity finite element model respectively; The average temperature and maximum temperature of the chip surface obtained by infrared camera testing and the finite element simulation calculation results are used to verify the correctness of the high-fidelity finite element model.

4. The high-voltage power module chip temperature inversion method according to claim 1, characterized in that: The construction of a multi-input and multi-output deep neural network specifically includes: A multi-layer perceptron structure is used as the network structure to build a multi-input and multi-output deep neural network, where the network input includes: several measurable point temperatures, and the network output includes several chip temperatures.

5. A high voltage power module chip temperature inversion system, characterized in that: include: A construction unit, used to construct a high-fidelity finite element model of the high-voltage power module to be inverted, and verify the correctness of the high-fidelity finite element model; A simulation unit, used to simulate based on the high-fidelity finite element model after correctness verification and according to preset boundary conditions, to obtain chip temperature and housing temperature for constructing a data set, wherein the preset boundary conditions include: different current levels and different heat sink temperatures; A training unit, used for building a multi-input multi-output deep neural network, and training the multi-input multi-output deep neural network through the preprocessed data set to obtain a deep neural network for characterizing the nonlinear mapping relationship between the measurable point temperature and the chip temperature; An acquisition unit is used to build a two-level inverter using a high-voltage power module and set the control algorithm and working parameters, and then control the load current by adjusting the load resistance, and simultaneously detect and obtain several groups of sample experimental data sets of chip temperature and shell temperature; The adjustment unit is used to fine-tune the parameters of the trained deep neural network based on the transfer learning theory and the sample experimental data set, so as to correct the nonlinear relationship and obtain the deep neural network finally used for inverting the chip temperature of the high-voltage power module.

6. The high-voltage power module chip temperature inversion system according to claim 5, characterized in that: The construction of a high-fidelity finite element model of the high-voltage power module to be inverted specifically includes: Obtaining the geometric dimensions of the high-voltage power module to be inverted and the material characteristic parameters of each component of the high-voltage power module to be inverted, wherein the geometric dimensions include: chip size, heat sink structure, welding layer thickness, and packaging material thickness; According to the geometric dimensions and the material characteristic parameters, a high-fidelity finite element model of the high-voltage power module to be inverted is constructed using finite element analysis software.

7. The high-voltage power module chip temperature inversion system according to claim 5, characterized in that: The verification of the correctness of the high-fidelity finite element model specifically includes: Using network units of different sizes to segment the chip, ceramic and copper substrate in the high-fidelity finite element model respectively; The average temperature and maximum temperature of the chip surface obtained by infrared camera testing and the finite element simulation calculation results are used to verify the correctness of the high-fidelity finite element model.

8. The high-voltage power module chip temperature inversion system according to claim 5, characterized in that: The construction of a multi-input and multi-output deep neural network specifically includes: A multi-layer perceptron structure is used as the network structure to build a multi-input and multi-output deep neural network, where the network input includes: several measurable point temperatures, and the network output includes several chip temperatures.

9. A high voltage power module chip temperature inversion device, characterized in that: The device comprises a processor and a memory: The memory is used to store program codes and transmit the program codes to the processor; The processor is used to execute the high-voltage power module chip temperature inversion method according to any one of claims 1 to 4 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store program codes, and the program codes are used to execute the high-voltage power module chip temperature inversion method according to any one of claims 1 to 4.

Images