Method and platform for evaluating aging performance of super junction mos device

Abstract

The application discloses a method and platform for evaluating aging performance of super junction MOS devices, and relates to the technical field of performance evaluation, comprising: performing test scene matching to obtain a target accelerated test scene group; calling a plurality of aging trend curves according to the target accelerated test scene group; locally calling a historical running data set of a super junction MOS device to be tested; extracting degradation features from the historical running data set to obtain a plurality of historical running time sequence segments; performing degradation increment correlation prediction on the plurality of historical running time sequence segments to output a device aging time sequence trend curve; comparing the device aging time sequence trend curve with a preset failure threshold to output a device remaining service life; and obtaining a device life design index of a target power electronic system through interaction, and combining the device remaining service life to quantitatively output an aging performance evaluation result. The application solves the technical problem of insufficient accuracy of aging performance evaluation in the prior art, and achieves the technical effect of improving the accuracy of aging performance evaluation.

Description

Technical Field

[0001] This invention relates to the field of performance evaluation technology, and more specifically to an aging performance evaluation method and platform for superjunction MOS devices. Background Technology

[0002] Superjunction MOS devices operate under high voltage and high temperature environments for extended periods, making them susceptible to electrothermal stress and subsequent performance degradation. Therefore, accurately assessing their aging performance is crucial for ensuring device reliability and predicting remaining lifespan. However, traditional aging performance assessment methods often rely on simple empirical data and incomplete models, failing to adequately consider variations in electrothermal stress under different application scenarios. Existing assessment methods also have limitations in extracting historical operating data and correlating aging trends, resulting in low accuracy in predictions and insufficient accuracy in assessing the aging performance of superjunction MOS devices. Summary of the Invention

[0003] This application provides a method and platform for evaluating the aging performance of superjunction MOS devices, which addresses the technical problem of insufficient accuracy in aging performance evaluation in the prior art.

[0004] In view of the above problems, this application provides a method and platform for evaluating the aging performance of superjunction MOS devices.

[0005] The first aspect of this application provides a method for evaluating the aging performance of a superjunction MOS device, the method comprising:

[0006] Test scenarios are matched based on the electrothermal stress characteristic parameters of the target application scenario of the superjunction MOS device under test to obtain a target accelerated test scenario group. Multiple aging trend curves are retrieved from the multi-scenario accelerated test database based on the target accelerated test scenario group, wherein the multiple aging trend curves are associated with multiple calibration operating condition parameter combinations. The historical operating dataset of the superjunction MOS device under test is locally retrieved. Degradation features are extracted from the historical operating dataset based on the multiple calibration operating condition parameter combinations to obtain multiple sets of historical operating sequence segments. Degradation increment correlation prediction is performed on the multiple sets of historical operating sequence segments based on the multiple aging trend curves, and the device aging time-series trend curve is output. The device's remaining service life is output by comparing the device aging time-series trend curve with a preset failure threshold. The equipment life design index of the target power electronic system is obtained interactively, wherein the superjunction MOS device under test is integrated into the target power electronic system. The aging performance evaluation result is quantitatively output based on the device's remaining service life and the equipment life design index.

[0007] A second aspect of this application provides an aging performance evaluation platform for superjunction MOS devices, the platform comprising:

[0008] The test scenario matching module is used to match test scenarios based on the electrothermal stress characteristic parameters of the target application scenario of the superjunction MOS device under test, and obtain a target accelerated test scenario group; the curve retrieval module is used to retrieve multiple aging trend curves from the multi-scenario accelerated test database based on the target accelerated test scenario group, wherein the multiple aging trend curves are associated with multiple calibration operating condition parameter combinations; the historical data retrieval module is used to locally retrieve the historical operating dataset of the superjunction MOS device under test; the feature extraction module is used to extract degradation features from the historical operating dataset based on the multiple calibration operating condition parameter combinations, and obtain multiple sets of historical operating sequence segments; the association prediction module is used to perform degradation increment association prediction of the multiple sets of historical operating sequence segments based on the multiple aging trend curves, and output the device aging time sequence trend curve; the comparison module is used to compare the device aging time sequence trend curve based on a preset failure threshold, and output the device remaining service life; the index acquisition module is used to interactively obtain the equipment life design index of the target power electronic system, wherein the superjunction MOS device under test is integrated into the target power electronic system; the evaluation module is used to quantitatively output the aging performance evaluation result based on the device remaining service life and the equipment life design index.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] This application performs test scenario matching based on the electrothermal stress characteristic parameters of the target application scenario of the superjunction MOS device under test, obtaining a target accelerated test scenario group; retrieves multiple aging trend curves from a multi-scenario accelerated test database based on the target accelerated test scenario group, wherein the multiple aging trend curves are associated with multiple calibration operating condition parameter combinations; locally calls the historical operating dataset of the superjunction MOS device under test; extracts degradation features from the historical operating dataset based on the multiple calibration operating condition parameter combinations, obtaining multiple sets of historical operating sequence segments; predicts the degradation increment of the multiple sets of historical operating sequence segments based on the multiple aging trend curves, and outputs the device aging time sequence trend curve; compares the device aging time sequence trend curve with a preset failure threshold, and outputs the remaining service life of the device; interactively obtains the equipment life design index of the target power electronic system, wherein the superjunction MOS device under test is integrated into the target power electronic system; and quantitatively outputs the aging performance evaluation result based on the remaining service life of the device and the equipment life design index. This invention addresses the technical problem of insufficient accuracy in aging performance assessment in existing technologies. By matching test scenarios based on electrothermal stress characteristic parameters and combining historical operating data with aging trend curves to predict degradation increments, it achieves the technical effect of improving the accuracy of aging performance assessment. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic diagram of the aging performance evaluation method for superjunction MOS devices provided in this application embodiment;

[0013] Figure 2 This is a schematic diagram of the aging performance evaluation platform for superjunction MOS devices provided in an embodiment of this application.

[0014] Figure labeling: Test scenario matching module 11, curve retrieval module 12, historical data retrieval module 13, feature extraction module 14, association prediction module 15, comparison module 16, indicator acquisition module 17, evaluation module 18. Detailed Implementation

[0015] This application provides an aging performance evaluation method and platform for superjunction MOS devices, addressing the technical problem of insufficient accuracy in aging performance evaluation in existing technologies. By matching test scenarios based on electrothermal stress characteristic parameters and combining historical operating data and aging trend curves to predict degradation increments, the technical effect of improving the accuracy of aging performance evaluation is achieved.

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0017] It should be noted that any variation of the terms "comprising" and "having" is intended to cover non-exclusive inclusion, for example, a process, method, platform, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such processes, methods, products, or devices.

[0018] Example 1, as Figure 1 As shown, this application provides a method for evaluating the aging performance of superjunction MOS devices, the method comprising:

[0019] Step S100: Perform test scenario matching based on the electrothermal stress characteristic parameters of the target application scenario of the superjunction MOS device under test to obtain the target accelerated test scenario group.

[0020] In this embodiment, test scenario matching is first performed based on the electrothermal stress characteristic parameters of the target application scenario of the superjunction MOS device under test. These electrothermal stress characteristic parameters refer to the stress characteristics formed by the combined action of electrical and thermal stresses during device operation. During test scenario matching, the electrothermal stress characteristic parameters are first analyzed and divided into electrical stress parameters and thermal stress parameters. Electrical stress parameters include the operating voltage range, switching frequency, effective current value, and peak current; thermal stress parameters include the average junction temperature, junction temperature fluctuation amplitude, and temperature change rate. Next, using the electrical stress parameters as the primary feature, electrical stress boundary thresholds are compared in a predefined scenario feature library to select multiple initial accelerated test scenario groups. Subsequently, using the thermal stress parameters as auxiliary features, secondary verification is performed on these initial accelerated test scenario groups to ensure they meet the requirements within the thermal stress range, ultimately selecting the target accelerated test scenario group that meets the criteria.

[0021] Furthermore, the method provided in the application embodiments, which matches test scenarios based on the electrothermal stress characteristic parameters of the target application scenario of the superjunction MOS device under test to obtain a target accelerated test scenario group, also includes:

[0022] The electrothermal stress characteristic parameters are analyzed to obtain electrical stress parameters and thermal stress parameters. The electrical stress parameters are used as the main features to perform electrical stress boundary threshold comparison in a predefined scene feature library to match multiple initial accelerated test scene groups. The thermal stress parameters are used as auxiliary features to perform a secondary verification of the thermal stress range compliance of the multiple initial accelerated test scene groups, and the target accelerated test scene group is selected and output.

[0023] Furthermore, the method provided in the application embodiments also includes:

[0024] The electrical stress parameters include the operating voltage range, switching frequency, effective current value, and peak current value; the thermal stress parameters include the average junction temperature, junction temperature fluctuation amplitude, and temperature change rate.

[0025] In this embodiment, the electrothermal stress characteristic parameters are first analyzed. The raw operating data of the superjunction MOS device under test in the target application scenario, including operating voltage, current, and junction temperature, are processed and categorized into electrical stress parameters and thermal stress parameters based on their stress sources. Electrical stress parameters include operating voltage range, switching frequency, RMS current, and peak current; thermal stress parameters include average junction temperature, junction temperature fluctuation amplitude, and temperature change rate. Through statistical analysis, peak value extraction, and feature quantification of the operating data, the impact of different stresses on the device aging process is clarified, thereby obtaining the electrical and thermal stress parameters.

[0026] Subsequently, the electrical stress parameters are used as the primary features, and an electrical stress boundary threshold comparison is performed in a predefined scene feature library. Specifically, the electrical stress boundary threshold information corresponding to each accelerated test scene in the predefined scene feature library is called, and the operating voltage range, switching frequency, effective current value, and peak current value are compared with the electrical stress threshold range in the scene for range inclusion and intersection determination. The comparison of the operating voltage range focuses on verifying whether the upper and lower limit ranges overlap, while the comparison of the switching frequency and current parameters is completed by determining whether they are within the preset threshold range. Test scenes that meet the electrical stress boundary conditions are retained to form multiple initial accelerated test scene groups.

[0027] Finally, thermal stress parameters are used as auxiliary features to perform a secondary screening of the initial accelerated test scenario group based on thermal stress range compliance. In this process, the average junction temperature, junction temperature fluctuation amplitude, and temperature change rate are compared one by one with the thermal stress threshold range in each initial accelerated test scenario to determine whether they are within the allowable thermal stress range. Test scenarios with thermal stress parameters exceeding the preset range are directly eliminated, retaining only scenarios whose thermal stress parameters fully meet the requirements of the target application scenario. Through this process, the screening range is tightened based on the already completed electrical stress matching, ensuring that the target accelerated test scenarios also have consistency in terms of thermal stress. The final target accelerated test scenario group is then output.

[0028] Step S200: Based on the target accelerated test scenario group, retrieve multiple aging trend curves from the multi-scenario accelerated test database, wherein the multiple aging trend curves are associated with multiple calibration operating condition parameter combinations.

[0029] Furthermore, the method provided in the application embodiments, which retrieves multiple aging trend curves from a multi-scenario accelerated testing database based on the target accelerated testing scenario group, further includes:

[0030] The system connects to the network and calls N calibrated operating condition parameter combinations of the first sample accelerated test scenario group; under the accelerated stress conditions corresponding to the target accelerated test scenario group, it uses the N calibrated operating condition parameter combinations to perform aging tests on the sample superjunction MOS device and collects N timing degradation data; it fits N aging trend curves based on the N timing degradation data; it binds the first sample accelerated test scenario group and the N aging trend curves and stores them in the multi-scenario accelerated test database; according to the target accelerated test scenario group, it retrieves the multiple aging trend curves by indexing the multi-scenario accelerated test database.

[0031] In this embodiment, when retrieving multiple aging trend curves from the multi-scenario accelerated test database based on the target accelerated test scenario group, the system first connects to the network and retrieves N calibration operating condition parameter combinations from the first sample accelerated test scenario group. These calibration operating condition parameter combinations represent different operating environment parameters, including operating voltage, switching frequency, RMS current, and junction temperature, among other electrothermal stress characteristics. These operating condition combinations are selected from a set of preset accelerated test scenarios according to the requirements of the target accelerated test scenario group, and are used to simulate various operating conditions that the device may encounter in actual applications. These operating condition combinations are automatically extracted from the multi-scenario accelerated test database via a network connection.

[0032] Next, under the accelerated stress conditions corresponding to the target accelerated test scenario group, the sample superjunction MOS devices were subjected to aging tests using the above N combinations of calibrated operating condition parameters. Each combination of operating conditions represents a different electrothermal stress environment. During the test, the devices were operated for a long time under these specific voltage, current, and temperature conditions. The current, voltage, temperature, and other performance data of the devices were monitored and recorded in real time to obtain N time-series degradation data.

[0033] Subsequently, based on N time-series degradation data, regression analysis methods, such as the least squares method, were used to fit N aging trend curves. Each aging trend curve corresponds to a set of calibration parameters, reflecting the performance degradation process of the device over time under a specific operating environment.

[0034] The first sample accelerated test scenario group is then bound to the generated N aging trend curves and stored in a multi-scenario accelerated test database. During this process, a database management system, such as a relational database management system, is used to associate and store the test scenarios with their corresponding aging trend curves, ensuring that each accelerated test scenario is closely linked to the corresponding aging data.

[0035] Finally, in the subsequent testing and performance evaluation process, based on the electrothermal stress characteristics of the target accelerated test scenario group, the system indexes and retrieves multiple aging trend curves that match the scenario from the multi-scenario accelerated test database.

[0036] Step S300: Locally access the historical running dataset of the superjunction MOS device under test.

[0037] In this embodiment, the historical operating dataset of the superjunction MOS device under test is retrieved locally from a preset database. The historical operating dataset includes information such as the device's historical operating voltage, switching frequency, current, and junction temperature.

[0038] Step S400: Based on the combination of multiple calibrated operating condition parameters, the degradation features of the historical operating dataset are extracted to obtain multiple sets of historical operating sequence fragments.

[0039] Furthermore, in the method provided in the application embodiments, the degradation feature extraction of the historical operation dataset based on the multiple combinations of calibrated operating condition parameters to obtain multiple sets of historical operation sequence fragments also includes:

[0040] Based on the relative stability of operating conditions, the historical operating dataset is decomposed to obtain M historical operating sequence segments; M operating condition attribute feature vectors of the M historical operating sequence segments are calculated; the multiple calibrated operating condition parameter combinations are traversed, the similarity matching of the M operating condition attribute feature vectors is performed, and the M historical operating sequence segments are divided into multiple sets of historical operating sequence segments corresponding to the multiple calibrated operating condition parameter combinations based on the nearest neighbor principle.

[0041] In this embodiment, when extracting degradation features from a historical operating dataset based on multiple combinations of calibrated operating condition parameters, the historical operating dataset is first decomposed into M historical operating sequence segments based on the relative stability of the operating conditions. During this process, a sliding window method is used to divide the historical operating dataset into fixed-size time periods in chronological order. For example, the window size is 1000 data points or each time period is 1 hour. Within each sliding window, the standard deviation of the data within that window is calculated, and a threshold is set to determine whether it represents a stable operating condition. For example, when the standard deviation is less than 0.02, the time segment is considered to represent a stable operating condition and is saved as a historical operating sequence segment. Through this method, the historical operating dataset is decomposed into M historical operating sequence segments according to the data stability standard.

[0042] Next, we calculate the M characteristic vectors of operating conditions for the M historical time segments. In this process, principal component analysis (PCA) is used to linearly transform multiple features in each historical time segment, such as operating voltage, switching frequency, RMS current, and junction temperature, to extract the most representative principal components, for example, selecting the first three principal components. The values ​​of these principal components form the characteristic vectors of operating conditions. Through this process, we obtain the M characteristic vectors of operating conditions.

[0043] Subsequently, multiple calibrated operating condition parameter combinations are traversed, and similarity matching is performed on M operating condition attribute feature vectors. In this process, the cosine similarity between the operating condition attribute feature vector of each historical time series segment and the operating condition attribute feature vector of the calibrated operating condition parameter combination is calculated using the cosine similarity method to evaluate their matching degree. The feature vectors of the historical time series segments are matched with the calibrated operating condition parameter combinations, and the operating condition parameter combination with the highest similarity is selected for classification. If the similarity is greater than a certain preset threshold, such as 0.95, the operating condition matching degree between the two is considered high.

[0044] Finally, based on the nearest neighbor principle, the M historical runtime time segments are divided into multiple time segments corresponding to multiple calibration operating condition parameter combinations. Using the K-Nearest Neighbor (KNN) algorithm, the K calibration operating condition parameter combinations most similar to the feature vector of each time segment are selected based on the calculated similarity; typically, K=5 is set. Through this process, based on the similarity between each historical time segment and the calibration operating condition combination, the time segment is assigned to the most matching operating condition group. Ultimately, the M historical runtime time segments are divided into multiple groups of historical runtime time segments, each group corresponding to a calibration operating condition parameter combination.

[0045] Step S500: Based on the multiple aging trend curves, perform degradation increment correlation prediction of the multiple sets of historical running timing segments, and output the device aging timing trend curve.

[0046] Furthermore, in the method provided in the application embodiments, the degradation increment correlation prediction of the multiple sets of historical runtime timing segments is performed based on the multiple aging trend curves, and the device aging timing trend curve is output, which further includes:

[0047] Based on the duration of the multiple sets of historical runtime timing segments, the multiple aging trend curves are quantitatively segmented to obtain multiple sets of degradation increment reference curve segments; based on the timestamp order mapping relationship between the multiple sets of historical runtime timing segments and M historical runtime timing segments, the multiple sets of degradation increment reference curve segments are reconstructed and spliced ​​to output the device aging reference timing curve; based on the device aging reference timing curve, degradation increment correlation prediction is performed to output the device aging timing trend curve.

[0048] In this embodiment, when predicting the degradation increment of multiple historical runtime segments based on multiple aging trend curves, the aging trend curves are first quantitatively segmented according to the duration of the multiple historical runtime segments. Each aging trend curve reflects the degradation process of the device under specific operating conditions. Based on the duration of each historical runtime segment, for example, each segment may represent 1 hour or 1000 data points, each aging trend curve is segmented according to these durations. This process uses a linear time segmentation method to divide each aging trend curve into equal time intervals, thereby obtaining multiple sets of degradation increment reference curve segments.

[0049] Next, based on the mapping relationship between multiple sets of historical runtime timing segments and the timestamp order of M historical runtime timing segments, the multiple sets of degradation increment reference curve segments are restored and stitched together. Each historical runtime timing segment has a timestamp indicating the time point of that segment. According to these timestamp order, the segmented degradation increment reference curve segments are stitched together in chronological order using a time alignment method to form the device aging reference timing curve.

[0050] Finally, degradation increment correlation prediction is performed based on the device aging baseline time series curve. In this process, firstly, M historical runtime time series segments are used as reference units, and M degradation increment values ​​are calculated from the device aging baseline time series curve. Then, the calibration stress intensity of multiple calibration operating condition parameter combinations is obtained interactively, and the actual stress intensity of the M historical runtime time series segments is locally retrieved. Based on the timestamp order mapping relationship of these time series segments, the calibration stress intensity and actual stress intensity are indexed, and M scaling factors are calculated. Finally, the degradation increment values ​​are scaled according to these scaling factors, and incremental accumulation fitting is performed, ultimately outputting the device aging time series trend curve.

[0051] Furthermore, in the method provided in the application embodiments, the degradation increment correlation prediction is performed based on the device aging reference timing curve, and the device aging timing trend curve is output, which further includes:

[0052] Using the M historical runtime timing segments as reference units, M degradation increment values ​​are calculated on the device aging reference timing curve; multiple calibration stress intensities of the multiple calibration operating condition parameter combinations are obtained interactively; M actual stress intensities of the M historical runtime timing segments are locally called; according to the timestamp order mapping relationship between the multiple sets of historical runtime timing segments and the M historical runtime timing segments, the multiple calibration stress intensities are indexed and allocated to the M actual stress intensities to calculate M scaling factors; after scaling the M degradation increment values ​​according to the M scaling factors, incremental accumulation fitting is performed, and the device aging timing trend curve is output.

[0053] In this embodiment, M historical timing segments are first used as reference units to calculate M degradation increment values ​​from the device aging reference timing curve. During this process, degradation data corresponding to each historical timing segment is extracted from the device aging reference timing curve to calculate the degradation increment value for each segment. The degradation increment value refers to the incremental portion of device performance change over time within the time period covered by each historical timing segment. This calculation process uses regression analysis or interpolation methods to find the performance change represented by each timing segment on the device aging reference timing curve, thereby deriving the degradation increment value for each segment.

[0054] Next, multiple calibration stress intensities are obtained through an interactive process, representing combinations of calibration operating condition parameters. These stress intensities are obtained through experiments or simulations based on preset calibration conditions, such as specific operating voltage, operating temperature, and load current, reflecting the stress intensity experienced by the device under different operating conditions.

[0055] Subsequently, the system locally retrieves the M actual stress intensities corresponding to the M historical runtime sequence segments. These actual stress intensity data originate from a real-time monitoring system or built-in sensors, recording parameters such as voltage, current, and temperature experienced by the device during actual use, reflecting the stress load the device bears during actual operation. By monitoring the actual operating conditions of the device, the actual stress intensity of each historical time segment is obtained.

[0056] Then, based on the mapping relationship between multiple sets of historical runtime sequence segments and the timestamp order of M historical runtime sequence segments, indexes are assigned to multiple calibrated stress intensities and M actual stress intensities. Each historical time sequence segment has a corresponding timestamp, and the calibrated stress intensities are mapped one-to-one with the actual stress intensities according to the order of these timestamps. Through this mapping, M scaling factors are calculated, which are the ratios of the actual stress intensity to the calibrated stress intensity for each historical time sequence segment.

[0057] Next, the M degradation increment values ​​are scaled using M scaling factors. Each degradation increment value is adjusted by multiplying it by the corresponding scaling factor. For example, if the degradation increment value of a historical time segment is 0.5, and the calculated scaling factor is 1.2, then the degradation increment value of that time segment will be scaled to 0.6. Finally, incremental accumulation fitting is performed. In this process, all degradation increment values ​​scaled by the scaling factors are accumulated, and a continuous device aging time-series trend curve is generated through regression analysis, such as linear regression or multinomial regression. The goal of this process is to accumulate the degradation increment values ​​of each time segment in chronological order to obtain the overall degradation amount of the device. The regression analysis fits a smooth curve based on the relationship between the accumulated degradation increment values ​​and time, reflecting the performance degradation trend of the device over a future period. Through these steps, the device aging time-series trend curve is generated.

[0058] Step S600: Based on the preset failure threshold, compare the device aging timeline trend curve and output the remaining service life of the device.

[0059] Furthermore, in the method provided in the application embodiments, the remaining service life of the device is output by comparing the device aging timeline trend curve with a preset failure threshold. This further includes:

[0060] Calculate the segmented degradation rate sequence of the device aging time-series trend curve; traverse the segmented degradation rate sequence based on a preset degradation rate threshold to locate acceleration inflection points; use the acceleration inflection points as degradation model switching points to extrapolate the device degradation trajectory segment by segment and output the predicted degradation trajectory; solve for the first intersection time between the predicted degradation trajectory and the preset failure threshold; use the time difference between the current timestamp and the first intersection time as the remaining service life of the device.

[0061] In this embodiment, when comparing the device aging time-series trend curve based on a preset failure threshold, the segmented degradation rate sequence of the device aging time-series trend curve is first calculated. In this process, the aging curve is divided into time periods, with each time period corresponding to the device's degradation rate within that period. The degradation rate refers to the amount of change in device performance per unit time, calculated as the ratio of the amount of performance degradation within that time period to the length of that time period. By performing this calculation for each time period, the degradation rate for each time period is obtained, forming a segmented degradation rate sequence.

[0062] Next, based on a preset degradation rate threshold, the segmented degradation rate sequence is traversed to locate the acceleration inflection point. During this process, a threshold comparison method is used to compare the degradation rate of each time segment with the preset degradation rate threshold. When the degradation rate of a certain time segment exceeds the preset degradation rate threshold, it indicates that the device has entered the accelerated degradation stage, and this moment is determined as the acceleration inflection point.

[0063] Subsequently, using the acceleration inflection point as the switching point for the degradation model, the device degradation trajectory is extrapolated in segments. The extrapolation method involves inferring future degradation trends based on known historical degradation data. Regression analysis, such as multinomial regression, is used to fit the data before the acceleration inflection point, thereby predicting the device's future aging trajectory during the accelerated degradation phase. This step yields the predicted degradation trajectory.

[0064] Next, the predicted degradation trajectory is compared with a preset failure threshold to determine the time of the first intersection. The preset failure threshold is a critical point representing an unacceptable performance degradation of the device, such as a performance drop to 50% of its original performance. By finding the intersection of the predicted degradation trajectory and the failure threshold, the device's failure state is determined. In this process, interpolation is used to calculate the intersection. Specifically, two data points close to the predicted degradation trajectory and the preset failure threshold are first identified. Then, linear interpolation or more complex interpolation methods, such as quadratic interpolation, are used between these two points to accurately calculate the intersection. This process yields the time of the first intersection.

[0065] Finally, the time difference between the current timestamp and the first intersection time is taken as the remaining lifetime of the device. The timestamp represents the current moment as recorded in real time, while the first intersection time is the calculated device failure time. By calculating the time difference between these two, the remaining lifetime of the device is obtained.

[0066] Step S700: Interact to obtain the device lifetime design specifications of the target power electronic system, wherein the superjunction MOS device to be tested is integrated into the target power electronic system.

[0067] In this embodiment, the device lifetime design specifications of the target power electronic system are obtained interactively. The target power electronic system refers to a complete power electronic system that integrates the superjunction MOS device to be tested. The device lifetime design specifications include key information such as the expected lifespan of the device, maximum junction temperature, and maximum load current.

[0068] Step S800: Quantitatively output the aging performance evaluation results based on the remaining service life of the device and the equipment life design indicators.

[0069] Furthermore, the method provided in the application embodiments, which quantifies and outputs aging performance evaluation results based on the remaining service life of the device and the equipment life design specifications, further includes:

[0070] Based on the current timestamp and the equipment lifespan design specifications, calculate the remaining service life of the equipment; calculate the lifespan deviation between the remaining service life of the equipment and the remaining service life of the components; normalize the lifespan deviation value to a relative health index, which is then output as the aging performance evaluation result.

[0071] In this embodiment, when determining the aging performance evaluation result, the remaining service life of the equipment is first calculated based on the current timestamp and the equipment life design specifications. Specifically, the usage time from the start of use to the current moment is first calculated based on the current timestamp. The usage time refers to the cumulative time since the equipment was put into use. Then, the remaining service life of the equipment is obtained by subtracting the usage time from the expected service life in the equipment life design specifications.

[0072] Next, the life deviation between the remaining service life of the equipment and the remaining service life of the components is calculated.

[0073] Next, the lifespan deviation values ​​are normalized using a minimum-maximum normalization method. The relative health index is obtained by dividing the difference between the minimum and maximum values ​​by the difference between the minimum and maximum values ​​of the lifespan deviation value. The minimum and maximum values ​​are the minimum and maximum values ​​of the lifespan deviation value in historical data. Finally, the calculated relative health index is output as the aging performance assessment result.

[0074] In summary, the embodiments of this application have at least the following technical effects:

[0075] This application performs test scenario matching based on the electrothermal stress characteristic parameters of the target application scenario of the superjunction MOS device under test, obtaining a target accelerated test scenario group; retrieves multiple aging trend curves from a multi-scenario accelerated test database based on the target accelerated test scenario group, wherein the multiple aging trend curves are associated with multiple calibration operating condition parameter combinations; locally calls the historical operating dataset of the superjunction MOS device under test; extracts degradation features from the historical operating dataset based on the multiple calibration operating condition parameter combinations, obtaining multiple sets of historical operating sequence segments; predicts the degradation increment of the multiple sets of historical operating sequence segments based on the multiple aging trend curves, and outputs the device aging time sequence trend curve; compares the device aging time sequence trend curve with a preset failure threshold, and outputs the remaining service life of the device; interactively obtains the equipment life design index of the target power electronic system, wherein the superjunction MOS device under test is integrated into the target power electronic system; and quantitatively outputs the aging performance evaluation result based on the remaining service life of the device and the equipment life design index. This invention addresses the technical problem of insufficient accuracy in aging performance assessment in existing technologies. By matching test scenarios based on electrothermal stress characteristic parameters and combining historical operating data with aging trend curves to predict degradation increments, it achieves the technical effect of improving the accuracy of aging performance assessment.

[0076] Example 2, based on the same inventive concept as the aging performance evaluation method for superjunction MOS devices in the foregoing examples, such as... Figure 2As shown, this application provides an aging performance evaluation platform for superjunction MOS devices. The platform and method embodiments in this application are based on the same inventive concept. The platform includes:

[0077] The test scenario matching module 11 is used to match test scenarios based on the electrothermal stress characteristic parameters of the target application scenario of the superjunction MOS device under test, and obtain a target accelerated test scenario group; the curve retrieval module 12 is used to retrieve multiple aging trend curves from the multi-scenario accelerated test database based on the target accelerated test scenario group, wherein the multiple aging trend curves are associated with multiple calibration operating condition parameter combinations; the historical data retrieval module 13 is used to locally retrieve the historical operating dataset of the superjunction MOS device under test; the feature extraction module 14 is used to extract degradation features from the historical operating dataset based on the multiple calibration operating condition parameter combinations. The system obtains multiple sets of historical runtime sequence segments; the correlation prediction module 15 is used to perform degradation increment correlation prediction of the multiple sets of historical runtime sequence segments based on the multiple aging trend curves, and outputs the device aging time sequence trend curve; the comparison module 16 is used to compare the device aging time sequence trend curve based on a preset failure threshold, and output the device's remaining service life; the index acquisition module 17 is used to interactively obtain the equipment life design index of the target power electronic system, wherein the superjunction MOS device under test is integrated into the target power electronic system; the evaluation module 18 is used to quantitatively output the aging performance evaluation result based on the device's remaining service life and the equipment life design index.

[0078] Furthermore, the platform is also used to implement the following functions:

[0079] The electrothermal stress characteristic parameters are analyzed to obtain electrical stress parameters and thermal stress parameters. The electrical stress parameters are used as the main features to perform electrical stress boundary threshold comparison in a predefined scene feature library to match multiple initial accelerated test scene groups. The thermal stress parameters are used as auxiliary features to perform a secondary verification of the thermal stress range compliance of the multiple initial accelerated test scene groups, and the target accelerated test scene group is selected and output.

[0080] Furthermore, the platform is also used to implement the following functions:

[0081] Based on the relative stability of operating conditions, the historical operating dataset is decomposed to obtain M historical operating sequence segments; M operating condition attribute feature vectors of the M historical operating sequence segments are calculated; the multiple calibrated operating condition parameter combinations are traversed, the similarity matching of the M operating condition attribute feature vectors is performed, and the M historical operating sequence segments are divided into multiple sets of historical operating sequence segments corresponding to the multiple calibrated operating condition parameter combinations based on the nearest neighbor principle.

[0082] Furthermore, the platform is also used to implement the following functions:

[0083] Based on the duration of the multiple sets of historical runtime timing segments, the multiple aging trend curves are quantitatively segmented to obtain multiple sets of degradation increment reference curve segments; based on the timestamp order mapping relationship between the multiple sets of historical runtime timing segments and M historical runtime timing segments, the multiple sets of degradation increment reference curve segments are reconstructed and spliced ​​to output the device aging reference timing curve; based on the device aging reference timing curve, degradation increment correlation prediction is performed to output the device aging timing trend curve.

[0084] Furthermore, the platform is also used to implement the following functions:

[0085] Using the M historical runtime timing segments as reference units, M degradation increment values ​​are calculated on the device aging reference timing curve; multiple calibration stress intensities of the multiple calibration operating condition parameter combinations are obtained interactively; M actual stress intensities of the M historical runtime timing segments are locally called; according to the timestamp order mapping relationship between the multiple sets of historical runtime timing segments and the M historical runtime timing segments, the multiple calibration stress intensities are indexed and allocated to the M actual stress intensities to calculate M scaling factors; after scaling the M degradation increment values ​​according to the M scaling factors, incremental accumulation fitting is performed, and the device aging timing trend curve is output.

[0086] Furthermore, the platform is also used to implement the following functions:

[0087] Calculate the segmented degradation rate sequence of the device aging time-series trend curve; traverse the segmented degradation rate sequence based on a preset degradation rate threshold to locate acceleration inflection points; use the acceleration inflection points as degradation model switching points to extrapolate the device degradation trajectory segment by segment and output the predicted degradation trajectory; solve for the first intersection time between the predicted degradation trajectory and the preset failure threshold; use the time difference between the current timestamp and the first intersection time as the remaining service life of the device.

[0088] Furthermore, the platform is also used to implement the following functions:

[0089] The system connects to the network and calls N calibrated operating condition parameter combinations of the first sample accelerated test scenario group; under the accelerated stress conditions corresponding to the target accelerated test scenario group, it uses the N calibrated operating condition parameter combinations to perform aging tests on the sample superjunction MOS device and collects N timing degradation data; it fits N aging trend curves based on the N timing degradation data; it binds the first sample accelerated test scenario group and the N aging trend curves and stores them in the multi-scenario accelerated test database; according to the target accelerated test scenario group, it retrieves the multiple aging trend curves by indexing the multi-scenario accelerated test database.

[0090] Furthermore, the platform is also used to implement the following functions:

[0091] Based on the current timestamp and the equipment lifespan design specifications, calculate the remaining service life of the equipment; calculate the lifespan deviation between the remaining service life of the equipment and the remaining service life of the components; normalize the lifespan deviation value to a relative health index, which is then output as the aging performance evaluation result.

[0092] Furthermore, the platform is also used to implement the following functions:

[0093] The electrical stress parameters include the operating voltage range, switching frequency, effective current value, and peak current value; the thermal stress parameters include the average junction temperature, junction temperature fluctuation amplitude, and temperature change rate.

[0094] It should be noted that the order of the embodiments described above is for descriptive purposes only and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. 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.

[0095] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for aging performance evaluation of super junction MOS devices, characterized in that, The method comprises: According to the electrical and thermal stress characteristic parameters of the target application scenario of the super-junction MOS device to be tested, a target accelerated test scenario group is obtained by matching the test scenario; According to the target accelerated test scenario group, a plurality of aging trend curves are retrieved from a multi-scenario accelerated test database, wherein the plurality of aging trend curves are associated with a plurality of calibrated operating condition parameter combinations; The historical running data set of the super-junction MOS device to be tested is locally called; According to the plurality of calibrated operating condition parameter combinations, degradation features are extracted from the historical running data set, and a plurality of historical running time sequence segments are obtained; According to the plurality of aging trend curves, degradation increment correlation prediction is performed on the plurality of historical running time sequence segments, and a device aging time sequence trend curve is output; Based on a preset failure threshold, the device aging time sequence trend curve is compared, and a device remaining useful life is output; The device life design index of the target power electronic system is obtained interactively, wherein the super-junction MOS device to be tested is integrated into the target power electronic system; According to the device remaining useful life and the device life design index, an aging performance evaluation result is quantitatively output.

2. The method for evaluating the burn-in performance of super junction MOS devices according to claim 1, wherein, According to the electrical and thermal stress characteristic parameters of the target application scenario of the super-junction MOS device to be tested, a target accelerated test scenario group is obtained by matching the test scenario, and the method comprises: The electrical and thermal stress characteristic parameters are analyzed to obtain electrical stress parameters and thermal stress parameters; The electrical stress parameters are taken as main features, and the electrical stress boundary threshold is compared in a pre-defined scene characteristic library to match a plurality of initial accelerated test scenario groups; The thermal stress parameters are taken as auxiliary features, and the plurality of initial accelerated test scenario groups are subjected to secondary verification of thermal stress range compliance to screen and output the target accelerated test scenario group.

3. The method for evaluating the burn-in performance of super junction MOS devices according to claim 1, wherein, According to the plurality of calibrated operating condition parameter combinations, degradation features are extracted from the historical running data set, and a plurality of historical running time sequence segments are obtained, and the method comprises: Based on the relative stability of the operating conditions, the historical running data set is decomposed to obtain M historical running time sequence segments; M operating condition attribute feature vectors of the M historical running time sequence segments are calculated; The plurality of calibrated operating condition parameter combinations are traversed, similarity matching of the M operating condition attribute feature vectors is performed, and the M historical running time sequence segments are divided into the plurality of historical running time sequence segments corresponding to the plurality of calibrated operating condition parameter combinations based on the nearest neighbor principle.

4. The method for evaluating the burn-in performance of super junction MOS devices according to claim 3, wherein, According to the plurality of aging trend curves, degradation increment correlation prediction is performed on the plurality of historical running time sequence segments, and a device aging time sequence trend curve is output, and the method comprises: According to the duration of the plurality of historical running time sequence segments, the plurality of aging trend curves are quantitatively segmented to obtain a plurality of degradation increment reference curve segments; According to the timestamp order mapping relationship between the plurality of historical running time sequence segments and the M historical running time sequence segments, the plurality of degradation increment reference curve segments are restored and spliced to output a device aging reference time sequence curve; Based on the device aging reference time sequence curve, degradation increment correlation prediction is performed to output the device aging time sequence trend curve.

5. The method for evaluating the burn-in performance of super junction MOS devices according to claim 4, wherein, The method comprises the following steps: Based on the device aging reference timing curve, degradation increment correlation prediction is performed, and a device aging timing trend curve is output. Taking the M historical running timing segments as reference units, M degradation increment values are calculated in the device aging reference timing curve. The multiple calibration stress intensities of the multiple calibration running condition parameter combinations are obtained interactively. The M actual stress intensities of the M historical running timing segments are called locally. According to the time stamp order mapping relationship between the multiple sets of historical running timing segments and the M historical running timing segments, the multiple calibration stress intensities are indexed to the M actual stress intensities to calculate M scaling factors.

6. The method for evaluating the burn-in performance of super junction MOS devices of claim 1, wherein, After the M scaling factors are used to scale the M degradation increment values, incremental accumulation fitting is performed, and the device aging timing trend curve is output. Based on the preset failure threshold, the device aging timing trend curve is compared, and a device remaining service life is output. The method comprises the following steps: The segmented degradation rate sequence of the device aging timing trend curve is calculated. Based on the preset degradation rate threshold, the segmented degradation rate sequence is traversed to locate an acceleration inflection point. Taking the acceleration inflection point as a degradation model switching point, the device degradation trajectory is segmented and extrapolated, and a predicted degradation trajectory is output.

7. The method for evaluating the burn-in performance of super junction MOS devices of claim 1, wherein, The first intersection time of the predicted degradation trajectory and the preset failure threshold is solved. The time difference between the current time stamp and the first intersection time is taken as the device remaining service life. According to the target accelerated test scene group, multiple aging trend curves are called from a multi-scene accelerated test database, and the method comprises the following steps: The N calibration running condition parameter combinations of the first sample accelerated test scene group are called in a networked manner. Under the accelerated stress conditions corresponding to the target accelerated test scene group, the N calibration running condition parameter combinations are used for aging test of a sample super-junction MOS device, and N timing degradation data are collected. Based on the N timing degradation data, N aging trend curves are fitted.

8. The method for evaluating the burn-in performance of super junction MOS devices according to claim 6, wherein, The first sample accelerated test scene group and the N aging trend curves are bound and stored in the multi-scene accelerated test database. According to the target accelerated test scene group, the multiple aging trend curves are indexed and called from the multi-scene accelerated test database. According to the device remaining service life and the equipment life design index, an aging performance evaluation result is quantitatively output, and the method comprises the following steps: According to the current time stamp and the equipment life design index, the equipment remaining service life is calculated.

9. The method for evaluating the burn-in performance of super junction MOS devices as claimed in claim 2, wherein, The life deviation value of the equipment remaining service life and the device remaining service life is calculated.

10. An aging performance evaluation platform for super junction MOS devices, characterized in that, The life deviation value is normalized as a relative health index, which is output as the aging performance evaluation result. The electrical stress parameters include working voltage range, switching frequency, current effective value and current peak value, and the thermal stress parameters include average junction temperature, junction temperature fluctuation amplitude and temperature change rate. The platform is used to perform the aging performance evaluation method for the super-junction MOS device as claimed in any one of claims 1-9, and the platform comprises: A test scene matching module is used to perform test scene matching according to the electrical and thermal stress characteristic parameters of a target application scene where a super-junction MOS device to be tested is located, so as to obtain a target accelerated test scene group. a curve retrieval module configured to retrieve a plurality of aging trend curves from a multi-scenario accelerated test database according to the target accelerated test scenario group, wherein the plurality of aging trend curves are associated with a plurality of calibrated operating condition parameter combinations; a historical data retrieval module configured to locally call a historical operating data set of the super-junction MOS device to be tested; a feature extraction module configured to extract degradation features from the historical operating data set according to the plurality of calibrated operating condition parameter combinations, to obtain a plurality of historical operating time sequence segments; a correlation prediction module configured to perform degradation increment correlation prediction on the plurality of historical operating time sequence segments according to the plurality of aging trend curves, to output a device aging time sequence trend curve; a comparison module configured to compare the device aging time sequence trend curve with a preset failure threshold, to output a device remaining useful life; an index acquisition module configured to interactively acquire a device life design index of a target power electronic system, wherein the super-junction MOS device to be tested is integrated into the target power electronic system; an evaluation module configured to quantitatively output an aging performance evaluation result according to the device remaining useful life and the device life design index.

Images