A method, system and device for monitoring operational faults of an IGBT

Abstract

The application relates to the technical field of IGBT fault testing, in particular to an IGBT operation fault monitoring method, system and device, which comprises the following steps: collecting collector-emitter voltage, collector current and IGBT shell temperature; determining periodicity retention and electrothermal risk index at the collection time; determining electrothermal risk deviation at the collection time according to the clustering result of the electrothermal risk index and the abnormal degree of the electrothermal risk index; and determining the adaptive value of the process noise variance of Kalman filtering according to the electrothermal risk deviation at the collection time, and obtaining the junction temperature of the IGBT by using Kalman filtering. The application can capture the junction temperature abnormity in time.

Description

Technical Field

[0001] This application relates to the field of IGBT fault testing technology, specifically to an IGBT operation fault monitoring method, system, and device. Background Technology

[0002] IGBT, or Insulated Gate Bipolar Transistor, is a core switching device in power electronic systems and is widely used in new energy power generation and smart grids. IGBTs operate under high voltage, high current, and high-frequency switching conditions for extended periods, making them extremely susceptible to overcurrent or overvoltage faults that could trigger power system shutdowns. Therefore, real-time monitoring of IGBT operating status is crucial.

[0003] The junction temperature of an IGBT chip during operation refers to the actual operating temperature of the PN junction inside the IGBT chip. It is also the highest temperature point during IGBT device operation and a core thermal characteristic parameter for measuring the IGBT's operating status and assessing device aging and failure risk. During operation, IGBTs generate power losses, causing the junction temperature to rise. However, junction temperature monitoring requires traversing the heat conduction path through the packaging material, resulting in a thermal hysteresis effect. Therefore, it is impossible to directly evaluate instantaneous changes in junction temperature based on the monitored temperature, making it difficult to promptly detect early junction temperature anomalies. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a method, system, and device for monitoring IGBT operational faults. The specific technical solution adopted is as follows: In a first aspect, embodiments of this application provide a method for monitoring operational faults of an IGBT, the method comprising the following steps: Collector-emitter voltage, collector current, and IGBT case temperature are collected within a sliding window at different acquisition times. Based on the periodicity of the collector current within a sliding window prior to the acquisition time, the periodicity retention rate at the acquisition time is determined. Combined with the degree to which the collector-emitter voltage, collector current, and IGBT case temperature are subjected to instantaneous interference from different components within a sliding window prior to the acquisition time, the electrothermal risk index of the IGBT at the acquisition time is determined. Based on the clustering results of the electrothermal risk index of all acquisition times within a sliding window before the acquisition time, and the degree of anomaly of the electrothermal risk index, the electrothermal risk deviation of the IGBT at the acquisition time is determined. The electrothermal risk deviation is used to characterize the significance of steady-state and transient deviations in junction temperature of the IGBT at the acquisition time. Based on the deviation of the electrothermal risk at the acquisition time, an adaptive value for the process noise variance of the Kalman filter is determined. The Kalman filter is then used to process the IGBT case temperature within a sliding window before the acquisition time to obtain the junction temperature of the IGBT at the acquisition time.

[0005] Furthermore, the process for determining the periodicity retention degree at the acquisition time is as follows: Based on the collector current collected within a sliding window prior to the acquisition time, establish the collector current sequence at the acquisition time. The maximum value of the autocorrelation coefficient of the collector current sequence at different time delays with a maximum time delay of the first preset duration is obtained and denoted as the periodicity retention degree at the acquisition time.

[0006] Furthermore, the process of constructing the electrothermal risk index at the time of data acquisition is as follows: Based on the degree to which the collector-emitter voltage, collector current, and IGBT case temperature are affected by different transient components within a sliding window before the acquisition time, the transient characteristic values ​​of the independent components at the acquisition time are determined. The electrothermal risk index of the IGBT at the acquisition time is determined. The electrothermal risk index is negatively correlated with the periodicity retention degree and positively correlated with the transient characteristic value.

[0007] Furthermore, the method for determining the transient eigenvalues ​​of the independent components at the acquisition time is as follows: Based on the collector-emitter voltage and IGBT case temperature acquired within a sliding window prior to the acquisition time, construct the collector-emitter voltage sequence and IGBT case temperature sequence at the acquisition time. Independent component analysis was performed on the collector-emitter voltage sequence, collector current sequence, and IGBT casing temperature sequence at the acquisition time to obtain three independent components. The kurtosis values ​​of the three independent components were recorded as the transient characteristic values ​​of the independent components at the acquisition time.

[0008] Furthermore, the process for determining the deviation of the electrothermal risk of the IGBT at the acquisition time is as follows: Based on the clustering results of the electrothermal risk index of all data collection times within a sliding window prior to the data collection time, a healthy state cluster, an early warning state cluster, and a fault state cluster are established. Obtain the LOF score of the electrothermal risk index for all sampling times within a sliding window prior to the sampling time; The difference between the electrothermal risk index at the acquisition time and the center of the healthy state cluster is recorded as the first difference at the acquisition time. The difference between the center of the fault state cluster and the center of the healthy state cluster is recorded as the second difference at the acquisition time. The ratio of the first difference to the second difference at the acquisition time is recorded as the steady-state deviation intensity at the acquisition time. The steady-state deviation intensity is used to characterize the degree of steady-state deviation of the junction temperature of the IGBT at the acquisition time. Based on the difference in LOF scores of the electrothermal risk index for all acquisition times within a sliding window prior to the acquisition time, the transient deviation intensity at the acquisition time is constructed. The transient deviation intensity is used to characterize the degree of transient deviation of the junction temperature of the IGBT at the acquisition time. The positive correlation between the steady-state deviation intensity and the transient deviation intensity at the acquisition time is denoted as the electrothermal risk deviation degree of the IGBT at the acquisition time.

[0009] Furthermore, the process for determining the health status cluster, early warning status cluster, and fault status cluster is as follows: Cluster the electrothermal risk index of all data collection times within a sliding window prior to the data collection time to obtain three clusters. According to the average electrothermal risk index contained in the clusters from smallest to largest, the corresponding clusters are successively named as the healthy state cluster, the early warning state cluster, and the fault state cluster.

[0010] Furthermore, the process for determining the transient deviation intensity at the acquisition moment is as follows: The ratio of the LOF score of the electrothermal risk index at the acquisition time to the maximum value of the LOF score of the electrothermal risk index at all acquisition times within the previous sliding window is denoted as the transient deviation intensity at the acquisition time.

[0011] Furthermore, the process for determining the adaptive value of the noise variance in the Kalman filter is as follows: The negative correlation processing result of the deviation of electrothermal risk at the acquisition time is recorded as the abnormal evaluation coefficient at the acquisition time. The product of the abnormal evaluation coefficient at the acquisition time and the initial process noise variance of the preset state variable is used as the value of the process noise variance of the Kalman filter.

[0012] Secondly, embodiments of this application also provide an IGBT operation fault monitoring device, the operation fault monitoring device including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0013] Thirdly, embodiments of this application provide an IGBT operation fault monitoring system, which includes: a data acquisition module, a transient interference evaluation module, a deviation degree evaluation module, and a junction temperature determination module.

[0014] The data acquisition module is used to acquire the collector-emitter voltage, collector current, and IGBT case temperature within a sliding window at different acquisition times. The transient interference evaluation module is used to determine the periodicity retention degree at the acquisition time based on the periodicity of the collector current in a sliding window before the acquisition time. Combined with the degree of transient interference of different components on the collector-emitter voltage, collector current and IGBT case temperature in a sliding window before the acquisition time, the module determines the electrothermal risk index of the IGBT at the acquisition time. The deviation evaluation module is used to determine the electrothermal risk deviation of the IGBT at the acquisition time based on the clustering results of the electrothermal risk index of all acquisition times within a sliding window before the acquisition time, and the degree of abnormality of the electrothermal risk index. The electrothermal risk deviation is used to characterize the significance of steady-state deviation and transient deviation of the junction temperature of the IGBT at the acquisition time. The junction temperature determination module is used to determine the adaptive value of the process noise variance of the Kalman filter based on the electrothermal risk deviation at the acquisition time. The Kalman filter is used to process the IGBT case temperature within a sliding window before the acquisition time to obtain the junction temperature of the IGBT at the acquisition time.

[0015] As can be seen from the above embodiments, the IGBT operation fault monitoring method, system, and device provided in this application have at least the following beneficial effects: This application first considers that voltage, current, and IGBT case temperature are interconnected and sensitive to noise. To avoid noise interference masking fault characteristics, based on the periodicity of the collector current and the degree of instantaneous interference from different components on different types of data, it evaluates the probability of current distortion or abnormal switching process within a sliding window before the acquisition time, and determines the electrothermal risk index of the IGBT at the acquisition time. Based on the clustering results of the electrothermal risk indices of all acquisition times within a sliding window before the acquisition time, it divides the IGBT into clusters consisting of electrothermal risk indices for healthy state, warning state, and fault state, and assigns an electrothermal risk index to each cluster. The central tendency of the data is determined, and the deviation between the operating state of the IGBT corresponding to the electrothermal risk index and its recent operating state is combined to evaluate the significance of steady-state and transient deviations of the junction temperature of the IGBT at the acquisition time, and obtain the electrothermal risk deviation degree of the IGBT at the acquisition time. Finally, based on the electrothermal risk deviation degree at the acquisition time, the adaptive value of the process noise variance of the Kalman filter is determined, and the Kalman filter is used to process the IGBT casing temperature within a sliding window before the acquisition time to obtain the junction temperature of the IGBT at the acquisition time. This solves the problem that the junction temperature is affected by the thermal hysteresis effect, making it difficult to capture junction temperature anomalies in a timely manner, and enables timely capture of junction temperature anomalies. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating the steps of an IGBT operation fault monitoring method provided in one embodiment of this application; Figure 2 This is a schematic diagram of the structure of an IGBT operation fault monitoring system provided in one embodiment of this application. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an IGBT operation fault monitoring method, system, and device proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] The following description, in conjunction with the accompanying drawings, details a specific scheme for an IGBT operation fault monitoring method, system, and equipment provided in this application.

[0020] Please see Figure 1 The diagram illustrates a flowchart of an IGBT operation fault monitoring method according to an embodiment of this application. The method includes the following steps: S001: Collector-emitter voltage, collector current, and IGBT case temperature within the sliding window at different acquisition times.

[0021] A high-voltage differential voltage sensor is connected in parallel between the collector and emitter power terminals of the IGBT to acquire the collector-emitter voltage. A high-frequency Hall current sensor is connected in series on the emitter circuit to acquire the collector current. A surface-mount temperature sensor is installed on the surface of the IGBT module housing to acquire the IGBT housing temperature.

[0022] Specifically, this embodiment uses a multi-channel high-speed ADC chip, sets the sampling frequency of collector-emitter voltage, collector current and IGBT case temperature to 100kHz, sets the sliding window length to 1s and the sliding step size to 10ms, sorts the collector-emitter voltage, collector current and IGBT case temperature collected in the sliding window before the acquisition time according to the order of acquisition, and obtains the collector-emitter voltage sequence, collector current sequence and IGBT case temperature sequence at the acquisition time.

[0023] Thus, the collector-emitter voltage sequence, collector current sequence, and IGBT case temperature sequence at each acquisition time have been obtained.

[0024] S002: Based on the periodicity of the collector current within a sliding window prior to the acquisition time, determine the periodicity retention rate at the acquisition time. Combined with the degree of transient interference of different components on the collector-emitter voltage, collector current, and IGBT casing temperature within a sliding window prior to the acquisition time, determine the electrothermal risk index of the IGBT at the acquisition time.

[0025] During operation, IGBTs generate power losses, which causes the junction temperature inside the IGBT to rise. However, the junction temperature cannot be measured directly, and indirect measurement methods are affected by thermal hysteresis, making the measured temperature unable to reflect the transient temperature fluctuations of the IGBT in a timely manner.

[0026] The collector current sequence at the acquisition time is processed using an autocorrelation analysis algorithm to obtain the autocorrelation coefficients under different time delays with a maximum time delay of the first preset duration. The maximum value of the autocorrelation coefficients under all time delays is recorded as the periodicity retention degree at the acquisition time.

[0027] In this embodiment, the value of the first preset duration is... The first preset duration is set to... It can cover the entire switching process of IGBT.

[0028] Autocorrelation analysis is commonly used to detect the periodicity and repetition of time series. The maximum value of the autocorrelation coefficient at different time delays is used to evaluate the strongest similarity between the time series and itself within the lag time. The larger the value, the stronger the periodicity of the time series. Therefore, the smaller the periodicity retention at the acquisition time, the less obvious the periodicity of the collector current acquired in the sliding window before the acquisition time, and the greater the possibility of current distortion or abnormal switching process in the sliding window before the acquisition time.

[0029] In the actual operation of IGBTs, voltage, current and IGBT case temperature are coupled with each other and are sensitive to noise. Based solely on the autocorrelation analysis of the collector current, it is difficult to distinguish between true electrical abnormalities, thermal abnormalities and noise interference. Noise interference can easily mask fault characteristics.

[0030] The FastICA independent component analysis algorithm based on maximum negative entropy was used to process the collector-emitter voltage sequence, collector current sequence, and IGBT case temperature sequence at the acquisition time. The number of independent components was set to 3, corresponding to the three independent characteristics of electrical characteristics, thermal characteristics, and noise characteristics, respectively. The three independent components and their kurtosis values ​​were obtained. The kurtosis values ​​of the three independent components were recorded as the transient characteristic values ​​of the independent components at the acquisition time.

[0031] Specifically, when using the FastICA independent component analysis algorithm, the maximum number of iterations is set to 500, and the convergence tolerance is [missing information]. To ensure separation stability; independent component analysis is a well-known technique and will not be elaborated further.

[0032] Independent component analysis (ICA) can separate independent signals from a mixed signal. The separated independent signals (specifically, the independent components corresponding to the independent signals) can characterize the interference of specific physical properties on the mixed signal. It's important to understand that the three independent components characterize electrical transient characteristics, thermal transient characteristics, and noise interference characteristics, respectively. The kurtosis value of the independent components reflects the sharpness of the distribution of their characteristics. The larger the kurtosis value, the more pronounced the peaks in the distribution of the independent component characteristics, thus enabling a more sensitive capture of the transient features of each independent component.

[0033] Based on the periodicity retention and transient characteristic value at the acquisition time, the electrothermal risk index of the IGBT at the acquisition time is determined. The electrothermal risk index at the acquisition time is negatively correlated with the periodicity retention at the acquisition time, and positively correlated with the transient characteristic value at the acquisition time.

[0034] It is understood that the positive and negative correlations in this application refer to the relationship between the independent and dependent variables. A positive correlation means that the dependent variable increases (decreases) as the independent variable increases (decreases), and can be an additive or multiplicative relationship. A negative correlation means that the dependent variable decreases (increases) as the independent variable increases (decreases), and can be an inverse relationship or a subtractive relationship.

[0035] Preferably, as an embodiment of this application, the negative of the periodicity retention at the acquisition time is used as the exponent of an exponential function with the natural constant as the base. The calculated value of the exponential function is recorded as the first exponential value at the acquisition time. The sum of all transient characteristic values ​​at the acquisition time is recorded as the first cumulative sum at the acquisition time. The preset first adjustment coefficient and second adjustment coefficient are used as the weights of the first exponential value and the first cumulative sum at the acquisition time, respectively. The first exponential value and the first cumulative sum at the acquisition time are weighted and summed. The result of the weighted summation is recorded as the electrothermal risk index of the IGBT at the acquisition time.

[0036] The sum of the preset first adjustment coefficient and the preset second adjustment coefficient is 1, and both the preset first adjustment coefficient and the preset second adjustment coefficient are greater than or equal to 0.3 and less than or equal to 0.7. In this embodiment, the value of the first adjustment coefficient and the second adjustment coefficient is 0.5.

[0037] It is important to understand that the first sum at the acquisition time is used to characterize the sum of the strength of all components' interference with IGBT operation. The higher the electrothermal risk index of the IGBT at the acquisition time, the greater the possibility of current distortion or abnormal switching process in the sliding window before the acquisition time.

[0038] Some other embodiments of this application may be that the ratio of the absolute deviation of the mean MAD of all transient characteristic values ​​at the acquisition time to the periodicity retention at the acquisition time is denoted as the electrothermal risk index of the IGBT at the acquisition time.

[0039] Thus, the electrothermal risk index of the IGBT at the time of data acquisition is obtained.

[0040] S003: Based on the clustering results of the electrothermal risk index of all acquisition times within a sliding window before the acquisition time, and the degree of anomaly of the electrothermal risk index, determine the electrothermal risk deviation of the IGBT at the acquisition time. The electrothermal risk deviation is used to characterize the significance of steady-state and transient deviations in the junction temperature of the IGBT at the acquisition time.

[0041] During the dynamic operation of IGBTs, the heat generated by the internal chip of the IGBT can only be transferred to the external sensor through multiple layers of packaging materials. Therefore, there is a significant time delay between the collected IGBT case temperature and the corresponding actual operating temperature. It is impossible to directly evaluate the instantaneous change of junction temperature based on the IGBT case temperature, and it is difficult to capture early junction temperature anomalies in a timely manner.

[0042] Cluster the electrothermal risk indexes of all data collection times within a sliding window prior to the data collection time, and set the number of clusters to 3. Obtain the 3 clusters and the center of the clusters. Calculate the mean of the electrothermal risk index contained in the clusters. According to the order of the mean of the electrothermal risk index from smallest to largest, the 3 clusters are successively named the healthy state cluster, the early warning state cluster, and the fault state cluster.

[0043] It is understandable that the cluster center of the cluster represents the central tendency of the electrothermal risk index within the cluster. The healthy state cluster, the early warning state cluster, and the fault state cluster correspond to the clusters composed of the electrothermal risk index under the healthy state, early warning state, and fault state of the IGBT, respectively.

[0044] The LOF (Local Outlier Factor) algorithm is used to process the electrothermal risk indexes of all acquisition times within a sliding window prior to the acquisition time, and the LOF score of each electrothermal risk index is obtained.

[0045] Understandably, the LOF score of the electrothermal risk index is used to evaluate the degree of anomaly of the electrothermal risk index relative to local nearest neighbors, that is, the degree of deviation between the operating state of the IGBT corresponding to the electrothermal risk index and its recent operating state. Therefore, the intermittent anomalies of the IGBT junction temperature process can be captured by the LOF score of the electrothermal risk index.

[0046] The difference between the electrothermal risk index at the acquisition time and the center of the healthy state cluster is recorded as the first difference at the acquisition time; the difference between the center of the fault state cluster and the center of the healthy state cluster is recorded as the second difference at the acquisition time; the ratio of the first difference to the second difference at the acquisition time is recorded as the steady-state deviation intensity at the acquisition time; the ratio of the LOF score of the electrothermal risk index at the acquisition time to the maximum value of the LOF score of the electrothermal risk index at all acquisition times within the previous sliding window is recorded as the transient deviation intensity at the acquisition time; the positive correlation result between the steady-state deviation intensity and the transient deviation intensity at the acquisition time is recorded as the electrothermal risk deviation degree of the IGBT at the acquisition time.

[0047] To avoid the denominator being zero during ratio calculation, a preset value needs to be added to the denominator. This preset value should be greater than or equal to... and less than or equal to The preset value is taken as follows in the example. .

[0048] It is understood that a positive correlation is applied to the steady-state and transient deviation intensities at the time of data acquisition, ensuring that both the steady-state and transient deviation intensities are positively correlated with the electrothermal risk deviation at the time of data acquisition. It is also understood that the positive correlation in this application refers to the relationship between the independent and dependent variables, where the independent variables are the steady-state and transient deviation intensities at the time of data acquisition, and the dependent variable is the electrothermal risk deviation at the time of data acquisition. A positive correlation means that the dependent variable increases (decreases) as the independent variable increases (decreases), and can be an additive or multiplicative relationship.

[0049] Preferably, as an embodiment of this application, the product of the steady-state deviation intensity and the transient deviation intensity at the acquisition time is denoted as the electrothermal risk deviation degree of the IGBT at the acquisition time.

[0050] The greater the deviation of the IGBT's electrothermal risk at the acquisition time, the more significant the steady-state and transient deviations of the IGBT's junction temperature at the acquisition time, and the greater the possibility of current distortion or abnormal switching process within a sliding window before the acquisition time.

[0051] Thus, the deviation of the electrothermal risk of the IGBT at the time of data acquisition is obtained.

[0052] S004: Based on the deviation of the electrothermal risk at the acquisition time, determine the adaptive value of the process noise variance of the Kalman filter, and use the Kalman filter to process the IGBT case temperature within a sliding window before the acquisition time to obtain the junction temperature of the IGBT at the acquisition time.

[0053] Thermal hysteresis affects the monitoring of junction temperature during IGBT operation, making it impossible to promptly identify transient temperature fluctuations in the chip. Therefore, an adaptive Kalman filter algorithm is employed to perform real-time filtering and state estimation on the IGBT case temperature data to predict the actual junction temperature change trend and alleviate the monitoring hysteresis problem caused by heat transfer delay. Specifically, based on the IGBT case temperature sequence and the IGBT case temperature at the acquisition time, combined with the junction temperature state estimation results from the previous adjacent acquisition time, the optimal estimate of the junction temperature at the acquisition time is obtained through Kalman filtering prediction and update steps, achieving real-time tracking of the junction temperature change trend.

[0054] In traditional Kalman filtering, the process noise variance is typically set to a fixed value, which can lead to filtering lag or over-smoothing. This makes it difficult to adapt to the dynamic operating conditions caused by electrothermal coupling during IGBT operation, and it fails to sensitively capture transient abnormal fluctuations in junction temperature, thus affecting monitoring accuracy. Therefore, an adaptive value for the process noise variance at the acquisition time is determined based on the electrothermal risk deviation of the IGBT at the acquisition time.

[0055] Preferably, as an embodiment of this application, the negative correlation processing result of the electrothermal risk deviation at the acquisition time is recorded as the abnormal evaluation coefficient at the acquisition time, and the product of the abnormal evaluation coefficient at the acquisition time and the initial process noise variance of the preset state variable is recorded as the process noise variance of the state variable at the acquisition time.

[0056] In this embodiment, the initial process noise variance of the preset state variable is set to a value of The unit of the initial process noise variance is the square of the temperature, and the unit of the initial process noise variance is the same as the unit of the variance of the state variables in the adaptive Kalman filter.

[0057] It is understood that a negative correlation processing is applied to the deviation of the electrothermal risk at the time of data collection, that is, to ensure that the deviation of the electrothermal risk at the time of data collection is negatively correlated with the anomaly evaluation coefficient at the time of data collection. It is understood that the negative correlation in this application refers to the relationship between the independent variable and the dependent variable, where the independent variable is the deviation of the electrothermal risk at the time of data collection, and the dependent variable is the anomaly evaluation coefficient at the time of data collection. The negative correlation means that the dependent variable decreases (increases) as the independent variable increases (decreases), and can be an inverse proportional relationship, a subtraction relationship, etc.

[0058] Preferably, as an embodiment of this application, the negative of the deviation of the electrothermal risk at the time of acquisition is taken as the exponent of an exponential function with the natural constant as the base, and the calculated value of the exponential function is recorded as the abnormality evaluation coefficient at the time of acquisition.

[0059] The process noise variance of the state variable at the acquisition time is determined based on the probability of current distortion or abnormal switching process within a sliding window prior to the acquisition time. This can reduce the value of the process noise variance when the IGBT operating state is abnormal, thereby decreasing the smoothness of the adaptive Kalman filter and improving its response speed to sudden changes. Furthermore, as the IGBT operating state approaches a healthy level, the process noise variance at the acquisition time can be made closer to the preset initial process noise variance value of the state variable, ensuring the smooth suppression effect of the adaptive Kalman filter on environmental noise and effectively mitigating the lag problem in junction temperature state estimation caused by thermal hysteresis.

[0060] The process noise variance is used to represent external disturbances in the statistical state equation, and its value directly determines the smoothness of the filter. It's important to note that since the state variable in a Kalman filter is the one-dimensional junction temperature, the process noise variance is a scalar value.

[0061] The process noise variance of the state variable at the acquisition time is used as the value of the process noise variance of the Kalman filter. The Kalman filter is used to process the IGBT case temperature at all acquisition times within a sliding window before the acquisition time to obtain the estimated value of the IGBT case temperature at the acquisition time.

[0062] The estimated value of the IGBT case temperature is the junction temperature of the IGBT.

[0063] This completes the operation monitoring of the IGBT.

[0064] Based on the same inventive concept as the above method, this application embodiment also provides an IGBT operation fault monitoring device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described IGBT operation fault monitoring methods.

[0065] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of an IGBT operation fault monitoring system according to one embodiment of this application. In this embodiment, the operation fault monitoring system includes units used to execute the steps in the corresponding embodiment of an IGBT operation fault monitoring method. See also... Figure 2 The fault monitoring system includes: a data acquisition module, a transient interference evaluation module, a deviation degree evaluation module, and a junction temperature determination module.

[0066] The data acquisition module is used to acquire the collector-emitter voltage, collector current, and IGBT case temperature within a sliding window at different acquisition times. The transient interference evaluation module is used to determine the periodicity retention degree at the acquisition time based on the periodicity of the collector current in a sliding window before the acquisition time. Combined with the degree of transient interference of different components on the collector-emitter voltage, collector current and IGBT case temperature in a sliding window before the acquisition time, the module determines the electrothermal risk index of the IGBT at the acquisition time. The deviation evaluation module is used to determine the electrothermal risk deviation of the IGBT at the acquisition time based on the clustering results of the electrothermal risk index of all acquisition times within a sliding window before the acquisition time, and the degree of abnormality of the electrothermal risk index. The electrothermal risk deviation is used to characterize the significance of steady-state deviation and transient deviation of the junction temperature of the IGBT at the acquisition time. The junction temperature determination module is used to determine the adaptive value of the process noise variance of the Kalman filter based on the electrothermal risk deviation at the acquisition time. The Kalman filter is used to process the IGBT case temperature within a sliding window before the acquisition time to obtain the junction temperature of the IGBT at the acquisition time.

[0067] It is understood that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0068] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0069] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the protection scope of this application.

Claims

1. A method for monitoring operational faults of an IGBT, characterized in that, The method includes the following steps: Collector-emitter voltage, collector current, and IGBT case temperature are collected within a sliding window at different acquisition times. Based on the periodicity of the collector current within a sliding window prior to the acquisition time, the periodicity retention rate at the acquisition time is determined. Combined with the degree to which the collector-emitter voltage, collector current, and IGBT case temperature are subjected to instantaneous interference from different components within a sliding window prior to the acquisition time, the electrothermal risk index of the IGBT at the acquisition time is determined. Based on the clustering results of the electrothermal risk index of all acquisition times within a sliding window before the acquisition time, and the degree of anomaly of the electrothermal risk index, the electrothermal risk deviation of the IGBT at the acquisition time is determined. The electrothermal risk deviation is used to characterize the significance of steady-state and transient deviations in junction temperature of the IGBT at the acquisition time. Based on the deviation of the electrothermal risk at the acquisition time, an adaptive value for the process noise variance of the Kalman filter is determined. The Kalman filter is then used to process the IGBT case temperature within a sliding window before the acquisition time to obtain the junction temperature of the IGBT at the acquisition time.

2. The IGBT operation fault monitoring method as described in claim 1, characterized in that, The process for determining the periodicity retention at the acquisition time is as follows: Based on the collector current collected within a sliding window prior to the acquisition time, establish the collector current sequence at the acquisition time. The maximum value of the autocorrelation coefficient of the collector current sequence at different time delays with a maximum time delay of the first preset duration is obtained and denoted as the periodicity retention degree at the acquisition time.

3. The IGBT operation fault monitoring method as described in claim 2, characterized in that, The process of constructing the electrothermal risk index at the time of data collection is as follows: Based on the degree to which the collector-emitter voltage, collector current, and IGBT case temperature are affected by different transient components within a sliding window before the acquisition time, the transient characteristic values ​​of the independent components at the acquisition time are determined. The electrothermal risk index of the IGBT at the acquisition time is determined. The electrothermal risk index is negatively correlated with the periodicity retention degree and positively correlated with the transient characteristic value.

4. The IGBT operation fault monitoring method as described in claim 3, characterized in that, The method for determining the transient eigenvalues ​​of the independent components at the acquisition time is as follows: Based on the collector-emitter voltage and IGBT case temperature acquired within a sliding window prior to the acquisition time, construct the collector-emitter voltage sequence and IGBT case temperature sequence at the acquisition time. Independent component analysis was performed on the collector-emitter voltage sequence, collector current sequence, and IGBT casing temperature sequence at the acquisition time to obtain three independent components. The kurtosis values ​​of the three independent components were recorded as the transient characteristic values ​​of the independent components at the acquisition time.

5. The IGBT operation fault monitoring method as described in claim 1, characterized in that, The process for determining the electrothermal risk deviation of the IGBT at the acquisition time is as follows: Based on the clustering results of the electrothermal risk index of all data collection times within a sliding window prior to the data collection time, a healthy state cluster, an early warning state cluster, and a fault state cluster are established. Obtain the LOF score of the electrothermal risk index for all sampling times within a sliding window prior to the sampling time; The difference between the electrothermal risk index at the acquisition time and the center of the healthy state cluster is recorded as the first difference at the acquisition time. The difference between the center of the fault state cluster and the center of the healthy state cluster is recorded as the second difference at the acquisition time. The ratio of the first difference to the second difference at the acquisition time is recorded as the steady-state deviation intensity at the acquisition time. The steady-state deviation intensity is used to characterize the degree of steady-state deviation of the junction temperature of the IGBT at the acquisition time. Based on the difference in LOF scores of the electrothermal risk index for all acquisition times within a sliding window prior to the acquisition time, the transient deviation intensity at the acquisition time is constructed. The transient deviation intensity is used to characterize the degree of transient deviation of the junction temperature of the IGBT at the acquisition time. The positive correlation between the steady-state deviation intensity and the transient deviation intensity at the acquisition time is denoted as the electrothermal risk deviation degree of the IGBT at the acquisition time.

6. The IGBT operation fault monitoring method as described in claim 5, characterized in that, The process for determining the health status cluster, early warning status cluster, and fault status cluster is as follows: Cluster the electrothermal risk index of all data collection times within a sliding window prior to the data collection time to obtain three clusters. According to the average electrothermal risk index contained in the clusters from smallest to largest, the corresponding clusters are successively named as the healthy state cluster, the early warning state cluster, and the fault state cluster.

7. The IGBT operation fault monitoring method as described in claim 5, characterized in that, The process for determining the transient deviation intensity at the acquisition moment is as follows: The ratio of the LOF score of the electrothermal risk index at the acquisition time to the maximum value of the LOF score of the electrothermal risk index at all acquisition times within the previous sliding window is denoted as the transient deviation intensity at the acquisition time.

8. The IGBT operation fault monitoring method as described in claim 1, characterized in that, The process for determining the adaptive value of the noise variance in the Kalman filtering process is as follows: The negative correlation processing result of the deviation of electrothermal risk at the acquisition time is recorded as the abnormal evaluation coefficient at the acquisition time. The product of the abnormal evaluation coefficient at the acquisition time and the initial process noise variance of the preset state variable is used as the value of the process noise variance of the Kalman filter.

9. An IGBT operation fault monitoring device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-8.

10. An IGBT operation fault monitoring system, implementing the method as described in claim 1, characterized in that, The operational fault monitoring system includes: The data acquisition module is used to acquire the collector-emitter voltage, collector current, and IGBT case temperature within a sliding window at different acquisition times. The transient interference evaluation module is used to determine the periodicity retention degree at the acquisition time based on the periodicity of the collector current in a sliding window before the acquisition time. Combined with the degree of transient interference of different components on the collector-emitter voltage, collector current and IGBT case temperature in a sliding window before the acquisition time, the module determines the electrothermal risk index of the IGBT at the acquisition time. The deviation evaluation module is used to determine the electrothermal risk deviation of the IGBT at the acquisition time based on the clustering results of the electrothermal risk index of all acquisition times within a sliding window before the acquisition time, and the degree of abnormality of the electrothermal risk index. The electrothermal risk deviation is used to characterize the significance of steady-state deviation and transient deviation of the junction temperature of the IGBT at the acquisition time. The junction temperature determination module is used to determine the adaptive value of the process noise variance of the Kalman filter based on the electrothermal risk deviation at the acquisition time. The Kalman filter is used to process the IGBT case temperature within a sliding window before the acquisition time to obtain the junction temperature of the IGBT at the acquisition time.

Images