An online aging state monitoring method and system for silicon carbide MOSFETs

Abstract

The present invention discloses an online aging state monitoring method and system for a silicon carbide MOSFET, belonging to the technical field of semiconductor device monitoring. Specifically, it includes: collecting the working voltage, working current and junction temperature data of the silicon carbide MOSFET in real time, analyzing and processing the data through a data processing unit, and extracting the aging state characteristic parameters of the MOSFET; using a dynamic aging state evaluation model that integrates multi-source data and time series information to evaluate the aging state of the silicon carbide MOSFET, and outputting the result to a display unit; setting a multi-level alarm strategy, triggering different levels of alarm signals according to the aging state. At the same time, collecting feedback data to optimize the state evaluation model, realizing real-time and accurate monitoring of the aging state of the silicon carbide MOSFET, and improving the reliability and stability of the power electronic system.

Description

Technical Field

[0001] The present invention belongs to the technical field of semiconductor device monitoring, and specifically relates to an online aging state monitoring method and system for silicon carbide MOSFETs. Background Art

[0002] Silicon carbide MOSFETs have been widely used in the field of power electronics, such as electric vehicles, photovoltaic inverters, and industrial motor drives, due to their advantages of high breakdown electric field, high electron mobility, and low on-resistance. However, during long-term operation, silicon carbide MOSFETs are affected by factors such as electrical stress and thermal stress, resulting in gradual degradation of their performance and aging phenomena. Aging may cause problems such as threshold voltage drift, increased on-resistance, and increased leakage current of the MOSFET, seriously affecting the reliability and stability of the power electronic system. Therefore, real-time and accurate monitoring of the aging state of silicon carbide MOSFETs is of great significance for ensuring the normal operation of the power electronic system.

[0003] Currently, the monitoring methods for the aging state of silicon carbide MOSFETs mainly include off-line detection and on-line detection. Off-line detection requires removing the device from the circuit and measuring parameters on specific test equipment. This method is cumbersome to operate and cannot monitor the aging of the device in actual operation in real time. Although on-line detection methods can monitor the device during operation, most of the existing on-line monitoring methods have problems such as single monitoring parameters, low accuracy, and complex algorithms, making it difficult to meet the requirements of actual applications.

[0004] For example, Chinese Patent No. CN118112386A discloses an online aging state monitoring method for silicon carbide MOSFETs, including: constructing an analysis model of the switching characteristics of silicon carbide MOSFETs that comprehensively considers package parasitic parameters; deriving the drain current increase rate according to the analysis model; defining a degradation precursor parameter based on the drain current increase rate, and monitoring whether the degradation precursor parameter is greater than zero: if yes, it is determined that the silicon carbide MOSFET has degraded, otherwise it is determined that the silicon carbide MOSFET is healthy. The beneficial effect of this technical solution is that an online aging state monitoring method for silicon carbide MOSFETs provided can comprehensively characterize two different degradation types, namely gate oxide degradation and bond wire fatigue, with only a single monitored quantity, reduce the complexity of the monitoring system, and improve the reliability of the state monitoring system.

[0005] The above existing technologies all have the following problems: Only relying on the single-dimensional monitored quantity of deriving the drain current increase rate and defining the degradation precursor parameter to determine the state of the silicon carbide MOSFET can only comprehensively characterize two degradation types, namely gate oxide degradation and bond wire fatigue, lacking comprehensive consideration of multiple factors; the applicable scenarios are relatively narrow, and there is a lack of multi-level early warning and intuitive display. Summary of the Invention

[0006] In view of the deficiencies of the prior art, the present invention proposes an online aging state monitoring method and system for silicon carbide MOSFETs, which can collect the working voltage, working current and junction temperature data of silicon carbide MOSFETs in real time, analyze and process the data through a data processing unit, and extract the characteristic parameters of the aging state of the MOSFETs; adopt a dynamic aging state evaluation model that integrates multi-source data and time series information to evaluate the aging state of silicon carbide MOSFETs, and output the results to a display unit; set a multi-level alarm strategy to trigger different levels of alarm signals according to the aging state, and at the same time, collect feedback data to optimize the state evaluation model, realizing real-time and accurate monitoring of the aging state of silicon carbide MOSFETs, and improving the reliability and stability of power electronic systems.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] An online aging state monitoring method for silicon carbide MOSFETs, comprising:

[0009] Step S1: When the silicon carbide MOSFET is working normally, use sensors to collect its working voltage, working current and junction temperature data in real time, and transmit the collected data to the data processing unit;

[0010] Step S2: The data processing unit analyzes and processes the received data, and extracts the characteristic parameters of the aging state of the MOSFET. The characteristic parameters of the aging state of the MOSFET include the change amount of the threshold voltage, the change rate of the on-resistance, the increase of the leakage current, and the change rate of the junction temperature;

[0011] Step S3: Based on the extracted characteristic parameters, use a pre-established dynamic aging state evaluation model that integrates multi-source data and time series information to evaluate the aging state of the silicon carbide MOSFET;

[0012] Step S4: According to the aging state evaluation result, output the aging state information of the silicon carbide MOSFET to the display unit for display. At the same time, set a multi-level alarm strategy, and automatically trigger different levels of alarm signals according to multiple dimensions such as the severity, change rate of the aging state, and potential impact on the system;

[0013] Step S5: Collect the feedback data of the aging state monitoring, and optimize the aging state evaluation model based on the feedback data.

[0014] Specifically, the extraction process of the characteristic parameters of the aging state of the MOSFET in step S2 includes: sending a specific test pulse signal to the silicon carbide MOSFET through the data processing unit, and measuring the gate-source voltage of the MOSFET using a voltage measurement circuit during the application of the test pulse and the drain-source voltage , meanwhile, set the voltage fluctuation threshold , if , then use the corresponding gate-source voltage as the threshold voltage of the MOSFET , then, use the threshold voltage of the MOSFET to calculate the difference with the initial threshold voltage of the MOSFET pre-stored in the data processing unit to obtain the threshold voltage change .

[0015] Specifically, the process of extracting the MOSFET aging state characteristic parameters in step S2 further includes: according to the real-time collected working current and working voltage data, calculate the real-time on-resistance of the MOSFET , and compare it with the initial on-resistance to obtain the on-resistance change rate ; real-time monitor the drain current of the MOSFET to obtain the drain current increment ; meanwhile, according to the collected junction temperature data , calculate the junction temperature change rate , where represents the initial drain current value, represents the initial junction temperature pre-stored in the data processing unit.

[0016] Specifically, the specific steps of step S3 include:

[0017] S3.1: Obtain the MOSFET aging state characteristic parameter set , and , and use to perform interpolation and supplementation on the obtained MOSFET aging state characteristic parameter set to obtain the interpolated and supplemented MOSFET aging state characteristic parameters , where represents the mth MOSFET aging state characteristic parameter, m represents the number of data in the MOSFET aging state characteristic parameter set, , , respectively represent the (i - 1)th, ith, and (i + 1)th MOSFET aging state characteristic parameters in X, , , respectively represent the time points corresponding to the (i - 1)th, ith, and (i + 1)th MOSFET aging state characteristic parameters;

[0018] S3.2: Read the characteristic parameter data within N sampling periods from the historical data storage area , and add the MOSFET aging state characteristic parameters after interpolation and supplementation as a new line to , generate a characteristic parameter data set, and then arrange the data in the characteristic parameter data set in chronological order to form a characteristic parameter sequence matrix .

[0019] Specifically, the specific steps of step S3 further include:

[0020] S3.3: Obtain the characteristic parameter sequence matrix , and perform vectorization processing to obtain a characteristic parameter vector ;

[0021] S3.4: Load the pre - constructed recurrent neural network model architecture from the model storage area and perform initialization operations;

[0022] S3.5: Input into the recurrent neural network model architecture to train the recurrent neural network model and obtain a dynamic aging state evaluation model;

[0023] S3.6: Input the real - time into the dynamic aging state evaluation model. The dynamic aging state evaluation model infers and predicts the aging state of the MOSFET and outputs an evaluation result vector representing the aging state .

[0024] Specifically, the specific steps of S4 include:

[0025] S4.1: Obtain the output result of the dynamic aging state evaluation model, organize the aging state information into a display data structure according to a predetermined format and layout, and transmit the aging state information to the display unit for real - time update and display through the established communication connection;

[0026] S4.2: After the display update is completed, judge whether the alarm condition is met according to the output result of the dynamic aging state evaluation model and the preset alarm thresholds at all levels;

[0027] If the dynamic aging state evaluation model determines that the MOSFET is in an unaged state and the aging state information is less than the first - level alarm threshold, the MOSFET state is normal, and it continues to wait for the update of the next evaluation result;

[0028] If the dynamic aging state evaluation model determines that the MOSFET is in a mildly aged state and the aging state information is greater than or equal to the first - level alarm threshold and less than the second - level alarm threshold, the MOSFET state is normal and triggers a first - level alarm, and regular inspections and maintenance are carried out;

[0029] If the dynamic aging state evaluation model determines that the MOSFET is in a moderately aged state, and the aging state information is greater than or equal to the secondary alarm threshold and less than the tertiary alarm threshold, then a secondary alarm is triggered, and monitoring measures are taken to adjust the system operating parameters;

[0030] If the dynamic aging state evaluation model determines that the MOSFET is in a severely aged state, and the aging state information is greater than or equal to the tertiary alarm threshold, then a tertiary alarm is triggered, and the device operation is immediately stopped and the MOSFET is replaced.

[0031] Specifically, the current aging state information of the MOSFET displayed in step S2 includes real-time characteristic parameter values, aging degree levels, and historical aging trend curves.

[0032] An on-line aging state monitoring system for a silicon carbide MOSFET, comprising: a data acquisition module, a data analysis module, an aging state evaluation module, a display and alarm module, and a model optimization module;

[0033] The data acquisition module is used to collect the working voltage, working current, and junction temperature data of the silicon carbide MOSFET in real time when the silicon carbide MOSFET is working normally;

[0034] The data analysis module is used to process and analyze the received data and extract the aging state characteristic parameters of the MOSFET;

[0035] The aging state evaluation module is used to evaluate the aging state of the silicon carbide MOSFET according to the extracted characteristic parameters;

[0036] The display and alarm module is used to output the aging state evaluation result to a display unit for display and trigger different levels of alarm signals according to the evaluation result;

[0037] The model optimization module is used to collect the feedback data of aging state monitoring and optimize the aging state evaluation model based on these data.

[0038] Specifically, the aging state evaluation module includes: a model selection unit and a state evaluation unit;

[0039] The model selection unit is used to construct and train a dynamic aging state evaluation model that integrates multi-source data and time series information;

[0040] The state evaluation unit is used to specifically evaluate the aging state of the MOSFET according to the application result of the model and obtain a quantitative index of the aging degree.

[0041] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of an online aging status monitoring method for silicon carbide MOSFET when executing the computer program.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention proposes an online aging status monitoring method for silicon carbide MOSFET, which realizes accurate evaluation of MOSFET aging status by real-time collection of working voltage, current and junction temperature data, and accurate extraction of aging status characteristic parameters, and then uses a dynamic evaluation model that integrates multi-source data and time series information. It can timely and accurately grasp the health status of the device, reduce the risk of system failure due to device aging, and improve the reliability and stability of the system.

[0044] 2. The present invention proposes an online aging status monitoring method for silicon carbide MOSFET, which intuitively displays the aging status information and sets a multi-level alarm strategy, so that the staff can quickly detect the abnormal aging of the device so as to take corresponding measures in time, such as maintenance and replacement of devices, to reduce potential economic losses; at the same time, the feedback data is used to optimize the evaluation model so that it can continuously adapt to the actual working conditions and aging characteristics changes of the MOSFET, further improving the accuracy and long-term effectiveness of the monitoring, and ensuring that reliable aging status monitoring services can be provided throughout the life cycle of the device. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A schematic diagram of an online aging status monitoring method for silicon carbide MOSFET according to the present invention;

[0046] Figure 2 This is a principle flow chart of an online aging status monitoring method for silicon carbide MOSFET of the present invention;

[0047] Figure 3 This is an architecture diagram of an online aging status monitoring system for silicon carbide MOSFET according to the present invention. DETAILED DESCRIPTION

[0048] Example 1

[0049] See also Figure 1 and Figure 2 , an embodiment of the present invention provides: an online aging status monitoring method for silicon carbide MOSFET, comprising the following steps:

[0050] Step S1: When the silicon carbide MOSFET is working normally, the operating voltage, operating current and junction temperature data thereof are collected in real time through the sensor, and the collected data are transmitted to the data processing unit;

[0051] Further, the specific steps of step S1 include:

[0052] (1) Perform power-on initialization operations on the sensor and its signal conditioning circuit, including setting the working parameters of the sensor, such as range and sensitivity, initializing the chip, configuring the amplifier gain and filter parameters in the signal conditioning circuit, etc., to ensure that the entire data acquisition system is in a normal working state and ready to collect data;

[0053] (2) Start the sensor to collect data according to the set sampling frequency. For example, if it is set to collect data every 1 millisecond, then every 1 millisecond, the voltage sensor, current sensor, and temperature sensor simultaneously measure the working voltage, working current, and junction temperature of the MOSFET;

[0054] (3) The sensor transmits the collected analog signals to their respective signal conditioning circuits. After amplification, filtering, and ADC conversion, digital data is obtained;

[0055] (4) The data is transmitted to the data processing unit through the selected data transmission interface. After receiving the data, the data processing unit stores it in the pre-opened data buffer and records the acquisition timestamp of the data, so that it can be corresponding to a specific time point during subsequent data processing and analysis, realizing real-time monitoring of the working state of the MOSFET;

[0056] (5) The data processing unit uses the parity check method to check the integrity and accuracy of the received data. If the check passes, it means the data is received normally and enters the next data processing process; if the check fails, a retransmission request is sent to the sensor to request retransmission of this group of data. At the same time, the data processing unit can count the number of error data. If there are multiple error data continuously, it means there is a fault in the data acquisition system, and fault diagnosis and elimination are required. For example, check whether the sensor connection is loose, whether the signal conditioning circuit is working properly, whether the data transmission line is interfered, etc., to ensure the reliability and stability of data acquisition. Among them, the parity check method is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here.

[0057] Step S2: The data processing unit analyzes and processes the received data, and extracts the MOSFET aging state characteristic parameters. The MOSFET aging state characteristic parameters include the threshold voltage change amount, on-resistance change rate, leakage current increment, and junction temperature change rate;

[0058] The current aging state information of the MOSFET shown in step S2 includes the real-time characteristic parameter values, aging degree levels, and historical aging trend curves.

[0059] Step S3: Based on the extracted feature parameters, use a pre-established dynamic aging state evaluation model that integrates multi-source data and time series information to evaluate the aging state of the silicon carbide MOSFET;

[0060] Step S4: According to the aging state evaluation result, output the aging state information of the silicon carbide MOSFET to the display unit for display. At the same time, set a multi-level alarm strategy, and automatically trigger alarm signals at different levels based on multiple dimensions such as the severity, change rate of the aging state, and potential impact on the system;

[0061] Step S5: Collect the feedback data of the aging state monitoring, and optimize the aging state evaluation model based on the feedback data.

[0062] Furthermore, the specific steps of Step S5 include:

[0063] (1) Establish a data collection channel to obtain feedback data related to the aging state monitoring of the silicon carbide MOSFET from multiple aspects. Among them, multiple aspects include: actual fault case data, recording detailed information such as the working conditions, characteristic parameter values, and fault types of the MOSFET when a fault occurs; regularly performed offline detection data, such as using professional test equipment to conduct a comprehensive performance test on the MOSFET to obtain accurate values of its precise threshold voltage, on-resistance, leakage current, etc. at specific time points; descriptions of abnormal phenomena and operation records feedback by users during actual use, such as abnormal heating, unstable power output, etc. that occur during system operation and the approximate working state of the MOSFET at that time;

[0064] (2) Sort and classify the collected feedback data, and conduct preliminary filing according to the type of data, such as fault data, detection data, user feedback data, time sequence, and correlation with the aging state of the MOSFET, for subsequent analysis and processing;

[0065] (3) Use a normalization method to preprocess the feedback data to make it match the data format and feature distribution used in the training of the dynamic aging state evaluation model. Among them, the normalization method is the prior art content in the field and is not the creative solution of this application, so it will not be elaborated here;

[0066] (4) Use the preprocessed feedback data to evaluate the existing dynamic aging state evaluation model to determine the performance and deficiencies of the model in actual applications. Among them, the feedback data is divided into two parts, one part is used as test data to directly evaluate the prediction accuracy of the model; the other part is used as analysis data to deeply explore the source of the model's prediction error and the distribution of feature importance;

[0067] (5) The dynamic aging state evaluation model is comprehensively evaluated using an accuracy evaluation index, and according to the evaluation results, the aging state evaluation model is updated and improved using regularization techniques. The regularization technique is the existing technical content in the field and is not the creative solution of this application, so it will not be elaborated here.

[0068] The process of extracting the aging state characteristic parameters of the MOSFET in step S2 includes: sending a specific test pulse signal to the silicon carbide MOSFET through the data processing unit. During the application of the test pulse, the gate-source voltage of the MOSFET is measured using a voltage measurement circuit. and the drain-source voltage , meanwhile, a voltage fluctuation threshold is set , if , then the corresponding gate-source voltage is used as the threshold voltage of the MOSFET , then, the threshold voltage of the MOSFET is subtracted from the initial threshold voltage of the MOSFET pre-stored in the data processing unit to calculate the threshold voltage change ;

[0069] According to the real-time collected working current and working voltage data, the real-time on-resistance of the MOSFET is calculated , and compared with the initial on-resistance to obtain the on-resistance change rate ; The drain current of the MOSFET is monitored in real time to obtain the drain current increment ; Meanwhile, according to the collected junction temperature data , the junction temperature change rate is calculated , where represents the initial drain current value, represents the initial junction temperature pre-stored in the data processing unit.

[0070] The specific steps of step S3 include:

[0071] S3.1: Obtain the aging state characteristic parameter set of the MOSFET , and , and use to interpolate and supplement the obtained aging state characteristic parameter set of the MOSFET to obtain the interpolated and supplemented aging state characteristic parameters of the MOSFET , where represents the m-th aging state characteristic parameter of the MOSFET, and m represents the number of data in the aging state characteristic parameter set of the MOSFET, , , respectively represent the aging state characteristic parameters of the (i - 1)-th, i-th, and (i + 1)-th MOSFETs in X. , , respectively represent the time points corresponding to the aging state characteristic parameters of the (i - 1)-th, i-th, and (i + 1)-th MOSFETs.

[0072] S3.2: Read the characteristic parameter data within N sampling periods from the historical data storage area , and use the interpolated and supplemented aging state characteristic parameters of the MOSFET as a new row and add it to , generate a characteristic parameter data set, and then arrange the data in the characteristic parameter data set in chronological order to form a characteristic parameter sequence matrix .

[0073] The specific steps of step S3 further include:

[0074] S3.3: Obtain the characteristic parameter sequence matrix , and perform vectorization processing on to obtain a characteristic parameter vector ;

[0075] S3.4: Load the pre - constructed recurrent neural network model architecture from the model storage area and perform an initialization operation. Among them, the recurrent neural network model architecture is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here;

[0076] S3.5: Input into the recurrent neural network model architecture to train the recurrent neural network model and obtain a dynamic aging state evaluation model. Among them, the training and prediction processes of the recurrent neural network model are the prior art content in this field and are not the creative solution of this application, so it will not be elaborated here;

[0077] S3.6: Input the real - time into the dynamic aging state evaluation model. The dynamic aging state evaluation model infers and predicts the aging state of the MOSFET and outputs an evaluation result vector representing the aging state .

[0078] Furthermore, the construction process of the dynamic aging state evaluation model includes:

[0079] (1) First, collect historical data of silicon carbide MOSFETs with different aging degrees, including operating voltages, currents, junction temperatures under various actual working conditions, and corresponding characteristic parameter data of aging states obtained through precise measurement and analysis. Then divide these data into a training set, a validation set, and a test set according to a certain ratio. The training set is used for model training, the validation set is used to verify and adjust the model's performance during training to prevent overfitting, and the test set is used to finally evaluate the model's generalization ability and accuracy.

[0080] (2) Then, select the recurrent neural network algorithm as the model's basic architecture, and design the specific structure of the model according to the selected basic architecture. Determine the number of neurons, connection methods, and various hyperparameters in the input layer, hidden layer, and output layer, such as the learning rate, batch size, and number of training epochs.

[0081] (3) In the model training stage, input the training set data into the model, calculate the output prediction value of the model through forward propagation, and then calculate the loss value using the cross-entropy loss function according to the difference between the prediction value and the true aging state label. At the same time, use the backpropagation algorithm to calculate the gradient of the model parameters with respect to the loss function, and update the model parameters according to the stochastic gradient descent optimization algorithm, continuously adjusting the weights and biases of the model, so that the model gradually learns the complex mapping relationship between the characteristic parameters and the aging state during training, thereby improving the model's prediction accuracy and generalization ability. During training, regularly use the validation set to verify the model, and adjust and optimize the model's hyperparameters according to the performance metrics on the validation set, such as accuracy, recall rate, and F1 value. When the performance of the model on the validation set no longer improves, stop training to prevent overfitting. Obtain the trained dynamic aging state evaluation model. It should be noted that the forward propagation algorithm, backpropagation algorithm, cross-entropy loss function, stochastic gradient descent optimization algorithm, and performance metric calculation formulas are all existing technical contents in this field and are not the creative solutions of this application, so they will not be elaborated here.

[0082] (4) Finally, input the real-time extracted and preprocessed characteristic parameters into the trained aging state evaluation model. The model infers and predicts the aging state of the MOSFET according to the patterns and rules it has learned, and outputs an evaluation result vector representing the aging state. This vector usually represents the probability distribution of the MOSFET belonging to each aging state category, such as normal, slightly aged, moderately aged, and severely aged.

[0083] The specific steps of S4 include:

[0084] S4.1: Obtain the output result of the dynamic aging state evaluation model, organize the aging state information into a display data structure according to a predetermined format and layout, and transmit the aging state information to the display unit through the established communication connection for real-time update and display;

[0085] S4.2: After the display update is completed, determine whether the alarm condition is met according to the output result of the dynamic aging state evaluation model and the preset alarm thresholds at all levels;

[0086] If the dynamic aging state evaluation model determines that the MOSFET is in the unaged state and the aging state information is less than the first-level alarm threshold, the MOSFET state is normal, and continue to wait for the update of the next evaluation result;

[0087] If the dynamic aging state evaluation model determines that the MOSFET is in the mildly aged state and the aging state information is greater than or equal to the first-level alarm threshold and less than the second-level alarm threshold, the MOSFET state is normal and triggers a first-level alarm, and regular inspections and maintenance are carried out;

[0088] If the dynamic aging state evaluation model determines that the MOSFET is in the moderately aged state and the aging state information is greater than or equal to the second-level alarm threshold and less than the third-level alarm threshold, trigger a second-level alarm and take monitoring measures to adjust the system operation parameters;

[0089] If the dynamic aging state evaluation model determines that the MOSFET is in the severely aged state and the aging state information is greater than or equal to the third-level alarm threshold, trigger a third-level alarm and immediately stop the device operation and replace the MOSFET.

[0090] Embodiment 2

[0091] Please refer to Figure 3 , another embodiment provided by the present invention: An on-line aging state monitoring system for a silicon carbide MOSFET, comprising:

[0092] A data acquisition module, a data analysis module, an aging state evaluation module, a display and alarm module, a model optimization module;

[0093] The data acquisition module is used to collect the working voltage, working current and junction temperature data of the silicon carbide MOSFET in real time when the silicon carbide MOSFET is working normally, providing a basis for subsequent data processing and analysis;

[0094] The data analysis module is used to process and analyze the received data and extract the aging state characteristic parameters of the MOSFET;

[0095] The aging state evaluation module is used to evaluate the aging state of the silicon carbide MOSFET according to the extracted characteristic parameters;

[0096] A display and alarm module, configured to output the aging state evaluation result to a display unit for display, and trigger alarm signals of different levels according to the evaluation result;

[0097] A model optimization module, configured to collect feedback data on aging state monitoring, and optimize the aging state evaluation model based on these data.

[0098] The data analysis module includes: a data receiving unit and a data analysis unit;

[0099] The data receiving unit is configured to receive data from the data acquisition module;

[0100] The data analysis unit is configured to process and analyze the received data, and extract aging state characteristic parameters of the MOSFET, such as the change amount of the threshold voltage, the change rate of the on-resistance, the increment of the leakage current, and the change rate of the junction temperature. These parameters can reflect the aging degree and state of the MOSFET.

[0101] The aging state evaluation module includes: a model selection unit and a state evaluation unit;

[0102] The model selection unit is configured to construct and train a dynamic aging state evaluation model that integrates multi-source data and time series information. This model can comprehensively consider various factors and accurately evaluate the aging state of the MOSFET;

[0103] The state evaluation unit is configured to specifically evaluate the aging state of the MOSFET according to the application result of the model, and obtain a quantitative index of the aging degree.

[0104] The display and alarm module includes: a display unit and an alarm strategy unit;

[0105] The display unit is configured to display the aging state information of the silicon carbide MOSFET, including the quantitative index of the aging degree and the characteristic parameters;

[0106] The alarm strategy unit, by setting a multi-level alarm strategy, automatically triggers alarm signals of different levels based on multiple dimensions such as the severity, change rate of the aging state, and potential impact on the system. These alarm signals can remind the operator to take timely measures to avoid the occurrence of faults.

[0107] Embodiment 3

[0108] An electronic device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it implements the steps of an online aging state monitoring method for a silicon carbide MOSFET. For details, refer to the above method embodiment and will not be elaborated here.

[0109] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0110] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0111] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments without departing from the spirit and scope of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. An on-line aging state monitoring method for a silicon carbide MOSFET, characterized in that Including: Step S1: When the silicon carbide MOSFET is operating normally, the working voltage, working current, and junction temperature data are collected in real time by sensors, and the collected data is transmitted to the data processing unit. Step S2: The data processing unit analyzes and processes the received data to extract the aging state characteristic parameters of the MOSFET. The aging state characteristic parameters of the MOSFET include the change amount of the threshold voltage, the change rate of the on-resistance, the increment of the leakage current, and the change rate of the junction temperature. The extraction process of the MOSFET aging state characteristic parameters in step S2 includes: sending a specific test pulse signal to the silicon carbide MOSFET through the data processing unit, and measuring the gate-source voltage of the MOSFET using a voltage measurement circuit during the application of the test pulse and the drain-source voltage . At the same time, set the voltage fluctuation threshold . If , then use the corresponding gate-source voltage as the threshold voltage of the MOSFET . Then, calculate the difference between the threshold voltage of the MOSFET and the initial threshold voltage of the MOSFET pre-stored in the data processing unit to obtain the threshold voltage change ; The process of extracting the characteristic parameters of the aging state of the MOSFET in step S2 further includes: according to the working current collected in real time and the working voltage data, calculate the real-time on-resistance of the MOSFET, and compare it with the initial on-resistance to obtain the on-resistance change rate ; monitor the drain current of the MOSFET in real time to obtain the drain current increment ; at the same time, according to the collected junction temperature data , calculate the junction temperature change rate , where represents the initial drain current value, represents the initial junction temperature pre-stored in the data processing unit; Step S3: Based on the extracted characteristic parameters, a dynamic aging state evaluation model that integrates multi-source data and time series information is used to evaluate the aging state of the silicon carbide MOSFET. S3.1: Obtain the set of characteristic parameters of the MOSFET aging state , and , and use to perform interpolation and supplementation on the obtained set of characteristic parameters of the MOSFET aging state to obtain the characteristic parameters of the MOSFET aging state after interpolation and supplementation , where represents the m-th characteristic parameter of the MOSFET aging state, and m represents the number of data in the set of characteristic parameters of the MOSFET aging state , , respectively represent the (i - 1)-th, i-th, and (i + 1)-th characteristic parameters of the MOSFET aging state in X , , respectively represent the time points corresponding to the (i - 1)-th, i-th, and (i + 1)-th characteristic parameters of the MOSFET aging state S3.2: Read the characteristic parameter data within N sampling periods from the historical data storage area , and use the MOSFET aging state characteristic parameters after interpolation and supplementation as a new row and add it to , generate a characteristic parameter data set, and then arrange the data in the characteristic parameter data set in chronological order to form a characteristic parameter sequence matrix ; Step S4: According to the aging state evaluation result, the aging state information of the silicon carbide MOSFET is output to the display unit for display. At the same time, a multi-level alarm strategy is set, and different levels of alarm signals are automatically triggered based on multiple dimensions such as the severity, change rate, and potential impact on the system of the aging state. Step S5: Collect the feedback data of the aging state monitoring, and optimize the aging state evaluation model based on the feedback data.

2. The online aging state monitoring method for a silicon carbide MOSFET according to claim 1, wherein The specific steps of step S3 further include: S3.3: Obtain the feature parameter sequence matrix , and perform vectorization processing to obtain the feature parameter vector ; S3.4: Load the pre-built recurrent neural network model architecture from the model storage area and perform initialization operations. S3.5: Input into the recurrent neural network model architecture to train the recurrent neural network model and obtain a dynamic aging state evaluation model; S3.6: Input the real-time into the dynamic aging state evaluation model. The dynamic aging state evaluation model infers and predicts the aging state of the MOSFET and outputs an evaluation result vector representing the aging state .

3. The on-line aging state monitoring method for a silicon carbide MOSFET according to claim 2, characterized in that, The specific steps of S4 include: S4.1: Obtain the output result of the dynamic aging state evaluation model, organize the aging state information into a display data structure according to a predetermined format and layout, and transmit the aging state information to the display unit for real-time update display through the established communication connection. S4.2: After the display update is completed, judge whether the alarm condition is met according to the output result of the dynamic aging state evaluation model and the preset alarm thresholds at all levels. If the dynamic aging state evaluation model determines that the MOSFET is in an unaged state and the aging state information is less than the first-level alarm threshold, the MOSFET state is normal, and it continues to wait for the update of the next evaluation result. If the dynamic aging state evaluation model determines that the MOSFET is in a mildly aged state and the aging state information is greater than or equal to the first-level alarm threshold and less than the second-level alarm threshold, the MOSFET state is normal, trigger a first-level alarm, and perform regular inspections and maintenance. If the dynamic aging state evaluation model determines that the MOSFET is in a moderately aged state and the aging state information is greater than or equal to the second-level alarm threshold and less than the third-level alarm threshold, trigger a second-level alarm, and take monitoring measures to adjust the system operation parameters. If the dynamic aging state evaluation model determines that the MOSFET is in a severely aged state and the aging state information is greater than or equal to the third-level alarm threshold, trigger a third-level alarm, and immediately stop the device operation and replace the MOSFET.

4. The on-line aging state monitoring method for a silicon carbide MOSFET according to claim 3, wherein, The current aging state information of the MOSFET displayed in step S2 includes the real-time characteristic parameter values, the aging degree level, and the historical aging trend curve.

5. An on-line aging state monitoring system for a silicon carbide MOSFET, which is used to implement the on-line aging state monitoring method for a silicon carbide MOSFET described in any one of claims 1-4, and is characterized in that, Including: Data acquisition module, data analysis module, aging state evaluation module, display and alarm module, model optimization module; The data acquisition module is used to collect the working voltage, working current and junction temperature data of the silicon carbide MOSFET in real time when the silicon carbide MOSFET is working normally; The data analysis module is used to process and analyze the received data and extract the aging state characteristic parameters of the MOSFET; The aging state evaluation module is used to evaluate the aging state of the silicon carbide MOSFET according to the extracted characteristic parameters; The display and alarm module is used to output the aging state evaluation result to the display unit for display and trigger alarm signals at different levels according to the evaluation result; The model optimization module is used to collect the feedback data of aging state monitoring and optimize the aging state evaluation model based on these data.

6. The on-line aging state monitoring system for a silicon carbide MOSFET according to claim 5, characterized in that, The aging state evaluation module includes: a model selection unit and a state evaluation unit; The model selection unit is used to construct and train a dynamic aging state evaluation model that integrates multi-source data and time series information; The state evaluation unit is used to specifically evaluate the aging state of the MOSFET according to the application results of the model and obtain a quantitative index of the aging degree.

7. An electronic device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of an on-line aging state monitoring method for a silicon carbide MOSFET according to any one of claims 1-4 are implemented.

Images