A neural network-based transformer junction temperature monitoring method and system

Abstract

The present invention discloses a method and system for monitoring junction temperature of a converter based on a neural network. j 、R dson 、R L 、R C The value of is preprocessed to complete the collection of the training data set of the maximum junction temperature model; the maximum junction temperature model based on the neural network is built; the input data set includes R dson 、R L 、R C ; The output data set is T j ; Use the training data set to train the maximum junction temperature model and obtain a parameter containing the independent variable R dson 、R L and R C And the dependent variable junction temperature data T j The relationship matrix of the converter is collected; the parameters of the inductor current, input voltage and capacitor voltage of the converter are collected; R is obtained by calculation dson 、R L and R C , R dson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j The present invention only needs to detect the inductor current, input voltage and output voltage of the converter to obtain the junction temperature parameters of the converter.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of junction temperature monitoring, and particularly relates to a transformer junction temperature monitoring method and system based on a neural network. BACKGROUND

[0002] SiCMOSFET is a new generation of devices for high switching frequency, high voltage and high temperature inverter. With the successful commercialization of wide bandgap devices, power applications based on SiCMOSFET are gradually increasing. So far, although SiC devices are designed to work at high temperatures, the maximum junction temperature is limited by the package and relatively small chip size. The junction temperature and temperature change play a key role in the reliable operation of power devices, and understanding the junction temperature of the device can achieve over-temperature protection and loss evaluation. In addition, the health status of the device can be monitored in real time through the junction temperature of the device, and various life extension strategies can be applied by adjusting the operation of the inverter accordingly. Since the field operating conditions of power converters are complex and unpredictable, the online junction temperature monitoring method of SiC devices is of great significance for safe operation and reliability evaluation. For relatively new SiCMOSFET with low on-resistance and fast switching, this feature is particularly valuable, because limited field data is available on its reliability, and the health status of the device needs to be monitored.

[0003] At present, one class of methods is based on the detection of semiconductor optical properties, which depend on temperature, or more simply, based on the acquisition of thermal images of semiconductor chips by infrared cameras. This type of technology can achieve high precision and provide thermal images of semiconductor chips, from which the highest temperature point and temperature gradient on the chip can be easily determined. However, all these methods require visual access to the chip, removal of the dielectric gel, and require additional computing resources to process thermal images. In addition, the physical contact method directly contacts the mold with a thermally sensitive material such as a thermocouple or thermistor. This method requires mechanical access to the mold inside the module, and has limited precision and dynamic response, which makes its application in commercial inverters impractical. Compared with direct methods using chip sensors or embedded package optical sensors, most junction temperature measurement methods based on electrical parameters are non-invasive and do not require modification of the device or package. This is very valuable for practical applications, because these simple plug-in circuits can be easily added or integrated into existing systems. SUMMARY

[0004] Therefore, in order to solve the above problems in the prior art, the present application provides a transformer junction temperature monitoring method and system based on a neural network, which only needs to detect the inductor current, input voltage and output voltage of the transformer to obtain the junction temperature parameters of the transformer.

[0005] The present application solves the above problems through the following technical means:

[0006] In a first aspect, the present invention provides a method for monitoring junction temperature of a converter based on a neural network, comprising the following steps:

[0007] When the switch tube of the converter is on, the expressions of the inductor current and capacitor voltage when the switch tube is on are obtained; when the switch tube of the converter is off, the expressions of the inductor current and capacitor voltage when the switch tube is off are obtained;

[0008] The inductor current, output voltage and input voltage of the converter are obtained in real time through the sampling circuit;

[0009] The parasitic resistance R of the switch tube is obtained by the expressions of the inductor current and capacitor voltage when the switch tube is turned on, the expressions of the inductor current and capacitor voltage when the switch tube is turned off, and the collected inductor current, input voltage and capacitor voltage. dson , the parasitic resistance of the inductor R L and the parasitic resistance R of the capacitor C ;

[0010] Measure the junction temperature data T of the converter at this time j ;

[0011] To maintain the stable operation of the converter, change the power level or input voltage level of the circuit to achieve different working conditions, and sample the input current, input voltage and capacitor voltage of different working conditions in real time; thus obtaining the parasitic resistance R of the switch tube under different working conditions dson , the parasitic resistance of the inductor R L , the parasitic resistance of the capacitor R C And the junction temperature data T j ;

[0012] For different T j 、R dson 、R L 、R C The value of is pre-processed to complete the collection of the training data set of the maximum junction temperature model;

[0013] Build a maximum junction temperature model based on neural network; input data set includes R dson 、R L 、R C Data; the output data set is T j ;

[0014] The maximum junction temperature model is trained using the training data set, and a parameter containing the independent variable R is obtained. dson 、R L and R C And the dependent variable junction temperature data T j The relational matrix of

[0015] Collect the parameters of the converter's inductor current, input voltage, and capacitor voltage; obtain R dson 、R L and R C The value of R dson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j .

[0016] Preferably, when the switch is turned on, the expressions of the inductor current and the capacitor voltage are:

[0017]

[0018]

[0019] When the switch is turned off, the expressions of the inductor current and capacitor voltage are:

[0020]

[0021]

[0022] Where L is the inductor, C is the capacitor, R is the output resistance, t is the time, V in is the input voltage, v C is the capacitor voltage, V o is the output voltage, i L is the inductor current, R dson 、R L and R C They are the parasitic resistances of the switch tube, inductor and capacitor respectively.

[0023] Preferably, the neural network selects a deep neural network as a data-driven learning method; the objective function used is the time and absolute error integral ITAE, and the optimal parameter combination is obtained by minimizing ITAE; ITAE is expressed as:

[0024] ITAE=∫t|e(t)|dt (5)

[0025] Where e(t) is the difference between the actual sample value and the expected value; ITAE is a cost function used as a model evaluation metric for regression models; the Gray Wolf Optimization Algorithm is selected as the optimizer of the cost function; and the deep neural network is trained using the Gray Wolf Optimization Algorithm.

[0026] Preferably, the neural network is a deep neural network, a convolutional neural network, a recursive neural network or a nonlinear autoregressive network.

[0027] In a second aspect, the present invention provides a converter junction temperature monitoring system based on a neural network, comprising the following steps:

[0028] The expression acquisition module is used to obtain the expressions of the inductor current and capacitor voltage when the switch tube of the converter is turned on; when the switch tube of the converter is turned off, the expression of the inductor current and capacitor voltage when the switch tube is turned off is obtained;

[0029] The current and voltage sampling module is used to obtain the inductor current, output voltage and input voltage of the converter in real time through the sampling circuit;

[0030] The parasitic resistance calculation module is used to obtain the parasitic resistance R of the switch tube through the expressions of the inductor current and capacitor voltage when the switch tube is turned on, the expressions of the inductor current and capacitor voltage when the switch tube is turned off, and the collected inductor current, input voltage and capacitor voltage. dson , the parasitic resistance of the inductor R L and the parasitic resistance R of the capacitor C ;

[0031] Junction temperature measurement module, used to measure the junction temperature data T of the converter at this time j ;

[0032] The working condition change module is used to maintain the stable operation of the converter, change the power level or input voltage level of the circuit, realize different working conditions, and sample the input current, input voltage and capacitor voltage of different working conditions in real time; thus, the parasitic resistance R of the switch tube under different working conditions is obtained. dson , the parasitic resistance of the inductor R L , the parasitic resistance R of the capacitor C And the junction temperature data T j ;

[0033] Data preprocessing module is used to process different T j 、R dson 、R L 、R C The value of is pre-processed to complete the collection of the training data set of the maximum junction temperature model;

[0034] Model building module, used to build a maximum junction temperature model based on neural network; input data sets include R dson 、R L 、R C Data; the output data set is T j ;

[0035] The model training module is used to train the maximum junction temperature model using the training data set and obtain a model containing the independent variable R dson 、R L and R C And the dependent variable junction temperature data Tj The relationship matrix of

[0036] The junction temperature output module is used to collect the parameters of the converter's inductor current, input voltage, and capacitor voltage; R dson 、R L and R C The value of R dson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j .

[0037] Preferably, when the switch is turned on, the expressions of the inductor current and the capacitor voltage are:

[0038]

[0039]

[0040] When the switch is turned off, the expressions of the inductor current and capacitor voltage are:

[0041]

[0042]

[0043] Where L is the inductor, C is the capacitor, R is the output resistance, t is the time, V in is the input voltage, v C is the capacitor voltage, V o is the output voltage, i L is the inductor current, R dson 、R L and R C They are the parasitic resistances of the switch tube, inductor and capacitor respectively.

[0044] Preferably, the neural network selects a deep neural network as a data-driven learning method; the objective function used is the time and absolute error integral ITAE, and the optimal parameter combination is obtained by minimizing ITAE; ITAE is expressed as:

[0045] ITAE=∫t|e(t)|dt (5)

[0046] Where e(t) is the difference between the actual sample value and the expected value; ITAE is a cost function used as a model evaluation metric for regression models; the Gray Wolf Optimization Algorithm is selected as the optimizer of the cost function; and the deep neural network is trained using the Gray Wolf Optimization Algorithm.

[0047] Preferably, the neural network is a deep neural network, a convolutional neural network, a recursive neural network or a nonlinear autoregressive network.

[0048] In a third aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the converter junction temperature monitoring method based on a neural network as described in the first aspect of the present invention are implemented.

[0049] In a fourth aspect, the present invention provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the converter junction temperature monitoring method based on a neural network as described in the first aspect of the present invention.

[0050] Compared with the prior art, the beneficial effects of the present invention include at least:

[0051] Compared with the traditional junction temperature monitoring method, the present invention does not require the addition of additional electrical parameter monitoring circuits. It only needs to detect the inductor current, input voltage and output voltage of the converter to obtain the junction temperature parameters of the converter, which can achieve high-precision monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0053] Figure 1 is an equivalent circuit diagram of the converter of the present invention;

[0054] Figure 2 is a flow chart of a method for monitoring junction temperature of a converter based on a neural network according to the present invention;

[0055] Figure 3 is the structural diagram of the selected neural network;

[0056] Figure 4 is an overview diagram of the converter junction temperature monitoring method based on a neural network of the present invention;

[0057] Figure 5 is a schematic diagram of a converter junction temperature monitoring system based on a neural network according to the present invention;

[0058] Figure 6 It is a structural block diagram of the electronic equipment of the present invention. DETAILED DESCRIPTION

[0059] To make the above-mentioned objectives, features, and advantages of the present invention more clearly understood, the technical solutions of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments described are only some of the embodiments of the present invention, not all of them. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are also within the scope of protection of the present invention.

[0060] Example 1

[0061] like Figure 2 As shown, the present invention provides a method for monitoring junction temperature of a converter based on a neural network, comprising the following steps:

[0062] S1. When the switch tube of the converter is on, obtain the expressions for the inductor current and capacitor voltage when the switch tube is on; when the switch tube of the converter is off, obtain the expressions for the inductor current and capacitor voltage when the switch tube is off;

[0063] S2, obtaining the inductor current, output voltage and input voltage of the converter in real time through the sampling circuit;

[0064] S3, the parasitic resistance R of the switch tube is obtained by the expressions of the inductor current and capacitor voltage when the switch tube is turned on, the expressions of the inductor current and capacitor voltage when the switch tube is turned off, and the collected inductor current, input voltage and capacitor voltage dson , the parasitic resistance of the inductor R L and the parasitic resistance R of the capacitor C ;

[0065] S4. Measure the junction temperature data T of the converter at this time j ;

[0066] S5. Maintain the stable operation of the converter, change the power level or input voltage level of the circuit, realize different working conditions, and sample the input current, input voltage and capacitor voltage of different working conditions in real time; thus, the parasitic resistance R of the switch tube under different working conditions is obtained. dson , the parasitic resistance of the inductor R L , the parasitic resistance R of the capacitor C And the junction temperature data T j ;

[0067] S6、For different T j 、R dson 、R L 、R C The value of is preprocessed to complete the collection of the training data set of the maximum junction temperature model;

[0068] S7. Build a maximum junction temperature model based on a neural network; the input data set includes R dson 、RL 、R C Data; the output data set is T j ;

[0069] S8. Use the training data set to train the maximum junction temperature model and obtain a model containing independent variables R dson 、R L and R C And the dependent variable junction temperature data T j The relationship matrix of

[0070] S9, collect the parameters of the converter's inductor current, input voltage, and capacitor voltage; obtain R dson 、R L and R C The value of R dson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j .

[0071] The following specifically introduces the converter junction temperature monitoring method based on neural network of the present invention.

[0072] like Figure 1 As shown in Figure 2, when the switch is turned on, the expressions of the inductor current and capacitor voltage are:

[0073]

[0074]

[0075] When the switch is turned off, the expressions of the inductor current and capacitor voltage are:

[0076]

[0077]

[0078] Where L is the inductor, C is the capacitor, R is the output resistance, t is the time, V in is the input voltage, v C is the capacitor voltage, V o is the output voltage, i L is the inductor current, R dson 、R L and R C They are the parasitic resistances of the switch tube, inductor and capacitor respectively.

[0079] Figure 2 The flowchart of the proposed junction temperature detection method is shown in Figure 2. Figure 2It can be seen that the inductor current, output voltage and input voltage can be obtained in real time through the sampling circuit. Then, the inductor current, input voltage and capacitor voltage are collected through equations (1)-(4) to obtain R dson 、R L and R C In addition, the junction temperature T of the switch tube is obtained by the optical fiber temperature sensor. j The value of T is used to form the target data in the modeling process. Then, the stable operation of the converter is maintained and the input current, input voltage and capacitor voltage are measured in real time. j 、R dson 、R L 、R C The values ​​are preprocessed to complete data collection.

[0080] A neural network is selected to train the maximum junction temperature model using a rich dataset obtained from dynamic steady-state calibration. The input dataset includes R dson 、R L 、R C Therefore, the output dataset is in accordance with T j Generated. Through the proposed process, from calibration to modeling, the actual junction temperature can be used as a new indicator. Furthermore, a maximum junction temperature model is constructed using a neural network model. Data-driven learning is performed using the collected dataset. In recent years, power electronics has actively considered machine learning algorithms such as deep neural networks, convolutional neural networks, recurrent neural networks, and nonlinear autoregressive neural networks (NARX).

[0081] Figure 3 : is a structural diagram of the selected neural network. Among them, the present invention selects a deep neural network as a data-driven learning method. The cost function is related to the error in the neural network. In particular, when implementing supervised learning, the cost function adjusts the weights and biases to reduce the error in the training data. When the value of the cost function decreases, the error of the designed neural network also decreases. The objective function used in the present invention is the integral of time and absolute error (ITAE), and the optimal parameter combination is obtained by minimizing ITAE. ITAE is expressed as:

[0082] ITAE=∫t|e(t)|dt (5)

[0083] Where e(t) is the difference between the actual sampled value and the expected value; ITAE is a cost function used as a model evaluation metric for regression models. The maximum junction temperature model is a regression problem; therefore, IATE is used as the cost function for deep neural networks. To minimize the cost function, a suitable intelligent optimization algorithm must be selected. In this paper, the Gray Wolf Optimization Algorithm (GWA) was selected as the optimizer for the cost function. Neural networks were trained using the GWA, and the results showed that this method has good performance.

[0084] The dataset, obtained through the neural network, contains experimental data. The collected data was randomly divided into 85% training data, 5% validation data, and 10% test data. The neural network learning process was implemented using the MATLAB Deep Learning Toolbox. The initial deep neural network structure and the optimized structure were studied on the dataset.

[0085] Figure 4 The schematic diagram of the proposed junction temperature detection method is shown in Figure 2. Maintaining stable operation of the converter to obtain R dson 、R L and R C And the junction temperature T is obtained in real time through the optical fiber method. j By training the neural network model in Matlab software with the data obtained in real time by the circuit and the optical fiber sensor, a neural network model containing the independent variable R is obtained. dson 、R L and R C And the independent variable junction temperature T j The data matrix.

[0086] In the experimental verification, the parameters of the converter's inductor current, input voltage, and capacitor voltage are collected. R dson 、R L and R C The collected value is input into the trained maximum junction temperature model to obtain the junction temperature data of the converter in real time.

[0087] Example 2

[0088] like Figure 5 As shown, the present invention provides a converter junction temperature monitoring system based on a neural network, comprising an expression acquisition module, a current and voltage sampling module, a parasitic resistance calculation module, a junction temperature measurement module, a working condition change module, a data preprocessing module, a model building module, a model training module and a junction temperature output module;

[0089] The expression acquisition module is used to obtain the expressions of the inductor current and capacitor voltage when the switch tube of the converter is turned on; when the switch tube of the converter is turned off, obtain the expressions of the inductor current and capacitor voltage when the switch tube is turned off;

[0090] The current and voltage sampling module is used to obtain the inductor current, output voltage and input voltage of the converter in real time through a sampling circuit;

[0091] The parasitic resistance calculation module is used to obtain the parasitic resistance R of the switch tube through the expressions of the inductor current and capacitor voltage when the switch tube is turned on, the expressions of the inductor current and capacitor voltage when the switch tube is turned off, and the collected inductor current, input voltage and capacitor voltage. dson , the parasitic resistance of the inductor R L and the parasitic resistance R of the capacitor C ;

[0092] The junction temperature measurement module is used to measure the junction temperature data T of the converter at this time. j ;

[0093] The working condition changing module is used to maintain the stable operation of the converter, change the power level or input voltage level of the circuit, realize different working conditions, and sample the input current, input voltage and capacitor voltage of different working conditions in real time; thereby obtaining the parasitic resistance R of the switch tube under different working conditions. dson , the parasitic resistance of the inductor R L , the parasitic resistance R of the capacitor C And the junction temperature data T j ;

[0094] The data preprocessing module is used to j 、R dson 、R L 、R C The value of is pre-processed to complete the collection of the training data set of the maximum junction temperature model;

[0095] The model building module is used to build a maximum junction temperature model based on a neural network; the input data set includes R dson 、R L 、R C Data; the output data set is T j ;

[0096] The model training module is used to train the maximum junction temperature model using the training data set and obtain a maximum junction temperature model containing the independent variable R dson 、R L and R C And the dependent variable junction temperature data T j The relationship matrix of

[0097] The junction temperature output module is used to collect the parameters of the converter's inductor current, input voltage and capacitor voltage; R dson 、R L and R C The value of Rdson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j .

[0098] The other features of this embodiment are the same as those of embodiment 1, so they will not be repeated here.

[0099] Example 3

[0100] Based on the same concept, the present invention also provides a schematic diagram of a physical structure, such as Figure 6 As shown, the server may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the steps of the neural network-based converter junction temperature monitoring method. For example, the steps include:

[0101] S1. When the switch tube of the converter is on, obtain the expressions for the inductor current and capacitor voltage when the switch tube is on; when the switch tube of the converter is off, obtain the expressions for the inductor current and capacitor voltage when the switch tube is off;

[0102] S2, obtaining the inductor current, output voltage and input voltage of the converter in real time through the sampling circuit;

[0103] S3, the parasitic resistance R of the switch tube is obtained by the expressions of the inductor current and capacitor voltage when the switch tube is turned on, the expressions of the inductor current and capacitor voltage when the switch tube is turned off, and the collected inductor current, input voltage and capacitor voltage dson , the parasitic resistance of the inductor R L and the parasitic resistance R of the capacitor C ;

[0104] S4. Measure the junction temperature data T of the converter at this time j ;

[0105] S5. Maintain the stable operation of the converter, change the power level or input voltage level of the circuit, realize different working conditions, and sample the input current, input voltage and capacitor voltage of different working conditions in real time; thus, the parasitic resistance R of the switch tube under different working conditions is obtained. dson , the parasitic resistance of the inductor R L , the parasitic resistance R of the capacitor C And the junction temperature data T j ;

[0106] S6、For different T j 、R dson 、R L 、R C The value of is preprocessed to complete the collection of the training data set of the maximum junction temperature model;

[0107] S7. Build a maximum junction temperature model based on a neural network; the input data set includes R dson 、R L 、R C Data; the output data set is T j ;

[0108] S8. Use the training data set to train the maximum junction temperature model and obtain a model containing independent variables R dson 、R L and R C And the dependent variable junction temperature data T j The relational matrix of

[0109] S9, collect the parameters of the converter's inductor current, input voltage, and capacitor voltage; obtain R dson 、R L and R C The value of R dson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j .

[0110] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the 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 and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program code.

[0111] Example 4

[0112] Based on the same concept, the present invention also provides a non-transitory computer-readable storage medium storing a computer program. The computer program includes at least one code segment that can be executed by a main control device to control the main control device to implement the steps of the neural network-based converter junction temperature monitoring method. For example, the steps include:

[0113] S1. When the switch tube of the converter is on, obtain the expressions for the inductor current and capacitor voltage when the switch tube is on; when the switch tube of the converter is off, obtain the expressions for the inductor current and capacitor voltage when the switch tube is off;

[0114] S2, obtaining the inductor current, output voltage and input voltage of the converter in real time through the sampling circuit;

[0115] S3, the parasitic resistance R of the switch tube is obtained by the expressions of the inductor current and capacitor voltage when the switch tube is turned on, the expressions of the inductor current and capacitor voltage when the switch tube is turned off, and the collected inductor current, input voltage and capacitor voltage dson , the parasitic resistance of the inductor R L and the parasitic resistance R of the capacitor C ;

[0116] S4. Measure the junction temperature data T of the converter at this time j ;

[0117] S5. Maintain the stable operation of the converter, change the power level or input voltage level of the circuit, realize different working conditions, and sample the input current, input voltage and capacitor voltage of different working conditions in real time; thus, the parasitic resistance R of the switch tube under different working conditions is obtained. dson , the parasitic resistance of the inductor R L , the parasitic resistance R of the capacitor C And the junction temperature data T j ;

[0118] S6、For different T j 、R dson 、R L 、R C The value of is pre-processed to complete the collection of the training data set of the maximum junction temperature model;

[0119] S7. Build a maximum junction temperature model based on a neural network; the input data set includes R dson 、R L 、R C Data; the output data set is T j ;

[0120] S8. Use the training data set to train the maximum junction temperature model and obtain a model containing independent variables R dson 、R Land R C And the dependent variable junction temperature data T j The relational matrix of

[0121] S9, collect the parameters of the converter's inductor current, input voltage, and capacitor voltage; obtain R dson 、R L and R C The value of R dson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j .

[0122] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in this application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive).

[0123] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

[0124] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A method for monitoring junction temperature of a converter based on a neural network, characterized in that: The steps include: When the switch tube of the converter is on, the expressions of the inductor current and capacitor voltage when the switch tube is on are obtained; when the switch tube of the converter is off, the expressions of the inductor current and capacitor voltage when the switch tube is off are obtained; The inductor current, output voltage and input voltage of the converter are obtained in real time through the sampling circuit; The parasitic resistance R of the switch tube is obtained by the expressions of the inductor current and capacitor voltage when the switch tube is turned on, the expressions of the inductor current and capacitor voltage when the switch tube is turned off, and the collected inductor current, input voltage and capacitor voltage. dson , the parasitic resistance of the inductor R L and the parasitic resistance R of the capacitor C ; Measure the junction temperature data T of the converter at this time j ; To maintain the stable operation of the converter, change the power level or input voltage level of the circuit to achieve different working conditions, and sample the input current, input voltage and capacitor voltage of different working conditions in real time; thus obtaining the parasitic resistance R of the switch tube under different working conditions dson , the parasitic resistance of the inductor R L , the parasitic resistance of the capacitor R C And the junction temperature data T j ; For different T j 、R dson 、R L 、R C The value of is preprocessed to complete the collection of the training data set of the maximum junction temperature model; Build a maximum junction temperature model based on neural network; input data set includes R dson 、R L 、R C Data; the output data set is T j ; The maximum junction temperature model is trained using the training data set, and a parameter containing the independent variable R is obtained. dson 、R L and R C And the dependent variable junction temperature data T j The relational matrix of Collect the parameters of the converter's inductor current, input voltage, and capacitor voltage; obtain R dson 、R L and R C The value of R dson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j .

2. The method for monitoring junction temperature of a converter based on a neural network according to claim 1, wherein: When the switch is turned on, the expressions of the inductor current and capacitor voltage are: When the switch is turned off, the expressions of the inductor current and capacitor voltage are: Where L is the inductor, C is the capacitor, R is the output resistance, t is the time, V in is the input voltage, v C is the capacitor voltage, V o is the output voltage, i L is the inductor current, R dson 、R L and R C They are the parasitic resistances of the switch tube, inductor and capacitor respectively.

3. The method for monitoring junction temperature of a converter based on a neural network according to claim 1, wherein: The neural network selects a deep neural network as a data-driven learning method; the objective function used is the time and absolute error integral ITAE, and the optimal parameter combination is obtained by minimizing ITAE; ITAE is expressed as: ITAE=∫t|e(t)|dt (5) Where e(t) is the difference between the actual sample value and the expected value; ITAE is a cost function used as a model evaluation metric for regression models; the Gray Wolf Optimization Algorithm is selected as the optimizer of the cost function; and the deep neural network is trained using the Gray Wolf Optimization Algorithm.

4. The method for monitoring junction temperature of a converter based on a neural network according to claim 1, wherein: The neural network is a deep neural network, a convolutional neural network, a recursive neural network or a nonlinear autoregressive network.

5. A converter junction temperature monitoring system based on a neural network, characterized in that: The steps include: The expression acquisition module is used to obtain the expressions of the inductor current and capacitor voltage when the switch tube of the converter is turned on; when the switch tube of the converter is turned off, the expression of the inductor current and capacitor voltage when the switch tube is turned off is obtained; The current and voltage sampling module is used to obtain the inductor current, output voltage and input voltage of the converter in real time through the sampling circuit; The parasitic resistance calculation module is used to obtain the parasitic resistance R of the switch tube through the expressions of the inductor current and capacitor voltage when the switch tube is turned on, the expressions of the inductor current and capacitor voltage when the switch tube is turned off, and the collected inductor current, input voltage and capacitor voltage. dson , the parasitic resistance of the inductor R L and the parasitic resistance R of the capacitor C ; Junction temperature measurement module, used to measure the junction temperature data T of the converter at this time j ; The working condition change module is used to maintain the stable operation of the converter, change the power level or input voltage level of the circuit, realize different working conditions, and sample the input current, input voltage and capacitor voltage of different working conditions in real time; thus, the parasitic resistance R of the switch tube under different working conditions is obtained. dson , the parasitic resistance of the inductor R L , the parasitic resistance of the capacitor R C And the junction temperature data T j ; Data preprocessing module is used to process different T j 、R dson 、R L 、R C The value of is preprocessed to complete the collection of the training data set of the maximum junction temperature model; Model building module, used to build a maximum junction temperature model based on neural network; input data sets include R dson 、R L 、R C Data; the output data set is T j ; The model training module is used to train the maximum junction temperature model using the training data set and obtain a parameter containing the independent variable R dson 、R L and R C And the dependent variable junction temperature data T j The relational matrix of The junction temperature output module is used to collect the parameters of the converter's inductor current, input voltage, and capacitor voltage; R dson 、R L and R C The value of R dson 、R L and R C The numerical value of is input into the maximum junction temperature model, and the junction temperature data T of the converter is obtained in real time based on the relational matrix. j .

6. The neural network-based converter junction temperature monitoring system according to claim 5, characterized in that: When the switch is turned on, the expressions of the inductor current and capacitor voltage are: When the switch is turned off, the expressions of the inductor current and capacitor voltage are: Where L is the inductor, C is the capacitor, R is the output resistance, t is the time, V in is the input voltage, v C is the capacitor voltage, V o is the output voltage, i L is the inductor current, R dson 、R L and R C They are the parasitic resistances of the switch tube, inductor and capacitor respectively.

7. The neural network-based converter junction temperature monitoring system according to claim 5, characterized in that: The neural network selects a deep neural network as a data-driven learning method; the objective function used is the time and absolute error integral ITAE, and the optimal parameter combination is obtained by minimizing ITAE; ITAE is expressed as: ITAE=∫t|e(t)|dt (5) Where e(t) is the difference between the actual sample value and the expected value; ITAE is a cost function used as a model evaluation metric for regression models; the Gray Wolf Optimization Algorithm is selected as the optimizer of the cost function; and the deep neural network is trained using the Gray Wolf Optimization Algorithm.

8. The neural network-based converter junction temperature monitoring system according to claim 5, characterized in that: The neural network is a deep neural network, a convolutional neural network, a recursive neural network or a nonlinear autoregressive network.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the converter junction temperature monitoring method based on a neural network as described in any one of claims 1 to 4 are implemented.

10. A non-transitory computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor, the steps of the converter junction temperature monitoring method based on a neural network are implemented as described in any one of claims 1 to 4.

Images