Dynamic Loss Detection Method and System for CoolMOS Devices

Abstract

The application provides a dynamic loss detection method and system for a CoolMOS device, relates to the technical field of semiconductor devices, and comprises the following steps: searching a historical use scene according to a carrying device ID and obtaining a plurality of standard test scenes through aggregation; performing correlation disturbance combination to obtain an intermediate state test scene array; after performing response delay test on the CoolMOS device in a to-be-detected carrying device, locally activating the intermediate state test scene array according to a test result, performing dynamic loss test on the CoolMOS device in the to-be-detected carrying device, and outputting an intermediate state multi-element loss track; and performing space-time evolution modeling and outputting a dynamic loss cloud chart. The application solves the technical problem that the prior art relies on fixed test methods and fixed test scenes and cannot comprehensively reflect the loss characteristics of the CoolMOS device under actual working conditions, resulting in insufficient authenticity and comprehensiveness of the loss detection result.

Description

Technical Field

[0001] This invention relates to the field of semiconductor device technology, and more specifically to a dynamic loss detection method and system for CoolMOS devices. Background Technology

[0002] CoolMOS devices possess low conduction losses and high switching efficiency, making them widely used in industrial and consumer electronics products. To ensure the efficient and stable operation of CoolMOS devices under extreme conditions such as high loads and high frequencies, evaluating their dynamic loss characteristics is crucial. However, existing technologies largely rely on fixed test methods and scenarios, only assessing the losses of the device under steady-state conditions. In real-world applications, devices often face dynamic conditions such as load variations, temperature fluctuations, and switching frequency changes. These dynamic changes significantly impact the device's losses. Fixed tests cannot fully reflect the loss characteristics of CoolMOS devices under actual operating conditions, resulting in insufficient accuracy and comprehensiveness of loss detection results, thus affecting system energy efficiency and long-term device reliability. Summary of the Invention

[0003] This application provides a dynamic loss detection method and system for CoolMOS devices, aiming to solve the technical problem that existing technologies rely heavily on fixed test methods and fixed test scenarios, which cannot fully reflect the loss characteristics of CoolMOS devices under actual working conditions, resulting in insufficient authenticity and comprehensiveness of loss detection results.

[0004] The first aspect disclosed in this application provides a dynamic loss detection method for CoolMOS devices. The method includes: retrieving historical usage scenarios based on the device ID, and obtaining multiple standard test scenarios by aggregating the historical usage scenarios; obtaining an intermediate state test scenario array by performing correlational perturbation combination on the multiple standard test scenarios; after performing response delay testing on the CoolMOS device in the device under test according to the multiple standard test scenarios, locally activating the intermediate state test scenario array according to the test results, performing dynamic loss testing on the CoolMOS device in the device under test, and outputting an intermediate state multi-dimensional loss trajectory; performing spatiotemporal evolution modeling on the intermediate state multi-dimensional loss trajectory, and outputting a dynamic loss cloud map.

[0005] The second aspect of this application discloses a dynamic loss detection system for CoolMOS devices. The system is used in the aforementioned dynamic loss detection method for CoolMOS devices. The system includes: a standard test scenario acquisition module, used to retrieve historical usage scenarios based on the device ID and obtain multiple standard test scenarios by aggregating the historical usage scenarios; a correlational perturbation combination module, used to obtain an intermediate-state test scenario array by performing correlational perturbation combination on the multiple standard test scenarios; a dynamic loss testing module, used to, after performing response delay testing on the CoolMOS device in the device under test according to the multiple standard test scenarios, locally activate the intermediate-state test scenario array based on the test results, perform dynamic loss testing on the CoolMOS device in the device under test, and output an intermediate-state multi-dimensional loss trajectory; and a spatiotemporal evolution modeling module, used to perform spatiotemporal evolution modeling on the intermediate-state multi-dimensional loss trajectory and output a dynamic loss cloud map.

[0006] One or more technical solutions provided in this application have at least the following beneficial effects:

[0007] By retrieving and aggregating historical usage scenarios, multiple standard test scenarios are obtained. These scenarios represent various working environments that the device may face in actual use. The aggregation process ensures the comprehensiveness and diversity of the test scenarios, enabling the tests to cover the device's performance under different conditions. By combining these standard test scenarios with correlated perturbations, an intermediate test scenario array is generated. These perturbation combinations simulate the dynamic changes of the device under various operating conditions, further increasing the diversity and complexity of the tests, thereby improving their realism and accuracy. By conducting response delay tests on the CoolMOS devices in the device under test based on these standard test scenarios, the response time of the CoolMOS devices under different operating conditions can be evaluated, providing the device with... Based on the test results, specific intermediate-state test scenario arrays are locally activated to assess latency performance under different loads. This local activation allows for refined testing of potential performance bottlenecks or critical scenarios, further outputting intermediate-state multi-dimensional loss trajectories. This makes the dynamic loss assessment more precise and realistic, enabling the identification of potential performance issues or optimization opportunities under specific conditions. By modeling the spatiotemporal evolution of the intermediate-state multi-dimensional loss trajectories, dynamic loss cloud maps can be generated by combining temporal and spatial changes. These dynamic loss cloud maps can intuitively display the loss evolution trend of the device under different operating conditions, helping engineers quickly identify high-loss regions of the device under different times and operating conditions, guiding subsequent optimization work, and thus effectively improving the energy efficiency and stability of CoolMOS devices.

[0008] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0009] Figure 1 This is a schematic flowchart of a dynamic loss detection method for CoolMOS devices provided in an embodiment of this application.

[0010] Figure 2 This is a schematic diagram of the dynamic loss detection system for CoolMOS devices provided in an embodiment of this application.

[0011] Figure labeling: Standard test scenario acquisition module 10, correlation disturbance combination module 20, dynamic loss test module 30, spatiotemporal evolution modeling module 40. Detailed Implementation

[0012] This application provides a dynamic loss detection method and system for CoolMOS devices, which solves the technical problem that existing technologies rely heavily on fixed test methods and fixed test scenarios, which cannot fully reflect the loss characteristics of CoolMOS devices under actual working conditions, resulting in insufficient authenticity and comprehensiveness of loss detection results.

[0013] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0014] Example 1, as Figure 1 As shown in the embodiments of this application, a dynamic loss detection method for CoolMOS devices is provided, the method comprising:

[0015] Historical usage scenarios are retrieved based on the device ID, and multiple standard test scenarios are obtained by aggregating these historical usage scenarios.

[0016] The device ID is used to distinguish different devices. By retrieving the device ID, historical data can be queried to identify the operating scenarios of that device under different usage conditions. These historical usage scenarios include different working environments, load conditions, temperature changes, etc. Aggregating these historical usage scenarios means organizing and classifying multiple historical usage scenarios to form representative scenarios with typical characteristics. These scenarios can be classified according to different operating conditions. The aggregation method can be based on cluster analysis based on operating condition characteristics, such as classifying scenarios according to the similarity of parameters such as temperature, load, and switching frequency. After aggregation, multiple standard test scenarios are obtained. The standard test scenarios represent the main possible operating modes of the device and serve as the benchmark for subsequent testing to simulate different working environments and usage conditions.

[0017] An intermediate test scenario array is obtained by combining the multiple standard test scenarios with correlated perturbations.

[0018] Correlated disturbances refer to perturbing each standard test scenario to a certain extent to generate new scenarios. These disturbances are aimed at non-ideal operating conditions that the equipment may encounter, such as temperature fluctuations, voltage fluctuations, and load changes. The disturbance methods include parameter disturbances, time disturbances, and mode disturbances. For example, making small changes to control parameters such as current, voltage, and temperature to simulate the equipment's response in extreme environments.

[0019] The perturbed standard test scenarios are combined, for example, by interpolating the perturbed standard test scenarios to generate multiple intermediate scenario sequences. This interpolation operation helps to maintain a smooth transition when switching scenarios and avoids drastic jumps or unnatural changes during the test process. The multiple intermediate scenario sequences are then aggregated to generate an intermediate test scenario array. This intermediate test scenario array is a perturbed version of multiple standard test scenarios, representing different working conditions, and is used for more detailed testing.

[0020] After performing response delay tests on the CoolMOS devices in the device under test according to the multiple standard test scenarios, the intermediate state test scenario array is locally activated based on the test results, and dynamic loss tests on the CoolMOS devices are performed on the device under test, outputting intermediate state multivariate loss trajectories.

[0021] Based on multiple standard test scenarios, response delay tests are conducted on the CoolMOS devices in the device under test. The goal of the response delay test is to measure the time delay from the change of the input signal to the change of the device output under specific standard test scenarios. This involves the switching response time of the testing device, the lag time from the input signal to the output result, etc.

[0022] Based on the test results, the intermediate state test scenario array is locally activated. This means that after the response delay test, the results determine which intermediate state test scenarios should be activated for further loss testing. Local activation refers to selecting a portion of the scenarios related to the current device state and response delay from the entire intermediate state test scenario array. These activated scenarios represent the device's operating state under specific delay conditions. Activating these scenarios allows for accurate evaluation of the dynamic loss characteristics of CoolMOS devices under different operating states in actual working conditions.

[0023] In the activated intermediate-state test scenario, dynamic loss testing of the CoolMOS device is performed on the device under test. This includes measuring the power loss, heat loss, and switching loss of the CoolMOS device under different operating conditions. During the test, the change in loss over time is monitored, especially the loss curves under different load conditions, operating frequencies, and temperature variations. Based on the test results, an intermediate-state multivariate loss trajectory is generated, reflecting the loss performance of the device under different scenarios.

[0024] Spatiotemporal evolution modeling is performed on the intermediate-state multivariate loss trajectory to output a dynamic loss cloud map.

[0025] Spatiotemporal evolution modeling of intermediate-state multivariate loss trajectories is performed. Spatiotemporal evolution refers to the process of loss changing with time and space. In the modeling process, techniques such as tensor modeling, time series analysis, and multidimensional regression analysis are used to reflect the evolution of loss under multiple variables. Based on the spatiotemporal evolution modeling results, a dynamic loss cloud map is output. The dynamic loss cloud map is an image that visualizes loss data in the spatiotemporal dimension. The loss data is presented as different cloud distributions, similar to a heatmap or 3D graphics. Each point or region in the map represents the loss value at a specific time and under specific conditions. The intensity of loss can be represented by color or different markers, which is used to intuitively display the loss status of the equipment under different operating conditions.

[0026] Furthermore, the method involves retrieving historical usage scenarios based on the device ID, and aggregating these historical usage scenarios to obtain multiple standard test scenarios.

[0027] Based on the device ID, operation logs of devices of the same model are collected to obtain a feature device log set; the feature device log set is segmented into scenarios based on physical event triggering to obtain multiple operation scenario slices; after extracting the operating condition feature vectors from the multiple operation scenario slices, operating condition aggregation is performed based on the extraction results to output the multiple standard test scenarios.

[0028] The system retrieves operating logs for devices of the same model based on their device IDs. These logs record the device's operation under different times and environmental conditions, including: operating parameters such as operating voltage, current, and load; device status such as whether it has entered overload or over-temperature states; and fault records such as whether any faults or warnings have occurred. Since devices of the same model have similar operating characteristics, selecting operating logs from these devices is more helpful in analyzing their common features. The resulting set of feature device logs contains operating records from multiple devices under various operating conditions, providing data support for subsequent analysis.

[0029] During equipment operation, different physical events trigger different operations or state changes. These physical events include temperature fluctuations, voltage fluctuations, load changes, and equipment failures. Each time a physical event occurs, the equipment's behavior and performance change. Based on these physical events, the characteristic equipment log set is segmented into scenarios. Scenario segmentation is determined by identifying the trigger points of physical events in the logs. When a physical event is detected, such as a sudden load change or excessive equipment temperature, the operating log is divided into different scenario slices based on the corresponding physical event. Each resulting operating scenario slice represents a specific working scenario in which the equipment experiences a series of similar working states. Through scenario segmentation, a more detailed analysis of the equipment's performance under different working conditions can be performed.

[0030] Operating condition feature vector extraction involves extracting features related to equipment performance from each slice of the operating scenario. These features include: electrical features, such as voltage, current, power, and frequency; thermal features, such as temperature changes and heat dissipation efficiency; and temporal features, such as duration and periodicity. Feature extraction employs machine learning methods to transform the complex log data of the equipment into simple numerical features, facilitating subsequent analysis.

[0031] Operating condition aggregation is based on the extracted operating condition feature vectors to aggregate different operating scenario slices. For example, based on similarity metrics such as Euclidean distance and cosine similarity, similar operating scenario slices are merged into a standard test scenario. The standard test scenario represents the typical performance of the equipment under specific conditions, providing a benchmark scenario for subsequent dynamic loss testing.

[0032] Furthermore, after extracting the working condition feature vectors from the multiple operating scenario slices, working condition aggregation is performed based on the extraction results to output the multiple standard test scenarios. The method includes:

[0033] A pre-constructed operating condition feature extraction model is used, wherein the operating condition feature extraction model has built-in parallel isolated electrical feature extraction channels, thermal dynamic feature extraction channels, and time feature extraction channels; a first operating scenario slice is loaded into the operating condition feature extraction model, and the first operating condition feature is analyzed and output, wherein the first operating condition feature consists of the effective value of current, the scale of current change, the amplitude of heat dissipation efficiency fluctuation, the average temperature rise rate, the duration of the scenario, and the median switching frequency; multiple operating scenario slices are input into the operating condition feature extraction model, and operating condition feature vectors are extracted by analogy to obtain multiple operating condition features; operating condition clustering based on recurrence density is performed on the multiple operating condition features to output the multiple standard test scenarios.

[0034] The operating condition feature extraction model is a multi-channel model used to extract operating condition features from slices of operating scenarios. These features indicate the performance of the equipment under different operating conditions and are used for subsequent dynamic loss testing. Specifically, the electrical feature extraction channel extracts features related to the equipment's electrical performance, such as current, voltage, power, and frequency. This channel extracts these data from the equipment's electrical signals and converts them into meaningful values. The thermal dynamic feature extraction channel handles features related to the equipment's thermal performance, including heat dissipation efficiency and temperature rise rate. It analyzes the temperature changes of the equipment under different loads and operating conditions and how these changes affect the overall performance of the equipment. The time feature extraction channel handles time-related features, such as the equipment's duty cycle, duration, and switching frequency, helping to analyze the equipment's time-series behavior, especially how time factors affect the equipment's dynamic loss under specific operating modes.

[0035] Parallel isolation means that the three channels are independent of each other in the working condition feature extraction model. Each channel focuses on extracting features relevant to its own domain and does not interfere with each other. This ensures that different types of features will not affect each other because they share the same data source, thereby ensuring the accuracy and reliability of each extracted feature.

[0036] The first operating scenario slice is any one of multiple operating scenario slices and is used as the current analysis object. The first operating scenario slice is loaded into the operating condition feature extraction model. Features related to each channel are extracted separately, and these features are aggregated to generate comprehensive first operating condition features. Among these features, the effective current value reflects the current intensity of the equipment in this scenario and is an important indicator in electrical characteristics; a higher effective value indicates a greater workload. The current variation scale describes the amplitude or rate of change of the current, representing the degree of current fluctuation during operation. A larger variation scale usually indicates unstable operating conditions or large load fluctuations. The heat dissipation efficiency fluctuation amplitude reflects the fluctuation of heat dissipation efficiency during operation; larger fluctuations indicate an unstable heat dissipation system, which may lead to overheating. The average temperature rise rate represents the rate at which the equipment temperature rises, typically higher under high load or high power output; it is a key feature for evaluating the thermal management performance of the equipment. The scenario duration indicates the duration of the scenario, helping to analyze the stability of the equipment under long-term operation; prolonged high loads may exacerbate equipment wear. The median switching frequency represents the median switching frequency of the equipment in this scenario; a higher frequency means the equipment is operating in a high-speed switching state, which has a greater impact on equipment wear and lifespan.

[0037] Similar to the first operating scenario slice, all multiple operating scenario slices are input into the operating condition feature extraction model and subjected to the same operating condition feature vector extraction process. Finally, multiple operating condition features are obtained. These operating condition features are the key performance indicators of the equipment in different scenarios, reflecting the operating status of the equipment under various operating conditions.

[0038] Recurrence density refers to the number of other data points surrounding a given data point within a certain neighborhood. A high recurrence density region indicates a high density of data points within that area, suggesting that the operating condition is frequently encountered during equipment operation. Operating condition clustering based on recurrence density involves aggregating operating condition features from high-density regions together to form a set of standard operating scenarios. This allows similar operating condition features to be grouped into the same class, enabling the use of representative standard test scenarios in subsequent testing. Through the clustering process, multiple standard test scenarios are ultimately output, each representing a typical operating state of the equipment under specific conditions, for use in subsequent dynamic wear testing. Operating condition features with high recurrence density indicate high frequency of occurrence during equipment operation, typically representing the equipment's typical operating state. Clustering based on recurrence density can identify and retain typical scenarios, thereby reducing the impact of redundant scenarios.

[0039] Furthermore, the method involves loading a slice of the first operating scenario into the operating condition feature extraction model and analyzing and outputting the first operating condition features.

[0040] After loading the first operating scenario slice into the operating condition feature extraction model: Step a: Extract the average test voltage through the voltage feature sampling unit in the electrical feature extraction channel, and extract the effective value of the current and the current change scale through the current feature sampling unit; Step b: Calculate the heat dissipation efficiency fluctuation amplitude and the average temperature rise rate through the thermal dynamic feature extraction channel; Step c: Calculate the scene duration and the median switching frequency through the time feature extraction channel; wherein, steps a, b, and c are executed in parallel and in isolation; the first operating condition features are obtained by structured storage of the effective value of the current, the current change scale, the heat dissipation efficiency fluctuation amplitude, the average temperature rise rate, the scene duration, and the median switching frequency.

[0041] After loading the first runtime scenario slice into the working condition feature extraction model, relevant features are extracted based on three pre-built channels, including:

[0042] In the electrical feature extraction channel, voltage data from the test is extracted through the voltage feature sampling unit. Voltage is one of the most critical electrical parameters during equipment operation. The average voltage value refers to the average value of the equipment's voltage signal over a specific time period, reflecting the stability of the equipment under that operating scenario. If the voltage fluctuates significantly, it means that the equipment is unstable. The effective current value is the standard measurement method for current signals, used to measure the actual effect of AC current. The effective current value of the equipment under that scenario is extracted through the current feature sampling unit. It represents the current intensity of the equipment under that operating condition and can reflect the load status of the equipment. A higher effective current value means a higher load or power output. The current change scale describes the amplitude and rate of change of the current signal, indicating the magnitude of current fluctuations under that scenario. A larger current change scale means instability or sudden changes in the equipment's workload.

[0043] Heat dissipation efficiency is an indicator reflecting the heat dissipation capacity of equipment. The fluctuation amplitude of heat dissipation efficiency indicates the degree of change in the heat dissipation efficiency of the equipment under the operating scenario. If the equipment is operating under high load, it will generate more heat, thereby affecting the heat dissipation efficiency. This feature is extracted by monitoring the temperature change of the equipment and the performance of the heat dissipation system. A large fluctuation amplitude indicates that the heat dissipation system is unstable under the operating condition, which may lead to the risk of overheating of the equipment. The average temperature rise rate refers to the rate at which the temperature of the equipment rises under the scenario. The faster the temperature rise rate, the more the equipment is operating under high load or high power output, which increases the thermal management pressure. A faster temperature rise rate may lead to thermal failure, affecting the reliability and lifespan of the equipment.

[0044] Scene duration refers to the length of time that a device maintains a certain state under a certain operating scenario. Long-term high-load operation will exacerbate the wear and tear on the device, especially increasing the requirements for the thermal management system. Scene duration is used to analyze the long-term stability of the device under this operating condition and reflects the performance of the device within the duration. The median switching frequency reflects the switching operation frequency of the device under this scenario. For devices involving switching operations, frequent switching operations usually lead to switching losses and additional energy consumption.

[0045] Steps a, b, and c involve electrical feature extraction, thermal dynamics feature extraction, and temporal feature extraction, respectively. To improve processing efficiency, these steps are executed in parallel, utilizing multiple feature extraction channels for simultaneous calculations instead of relying on sequential execution. This improves processing speed, and isolated execution means that each step does not affect the others during execution. The extraction of electrical, thermal dynamics, and temporal features is independent and does not interfere with each other, ensuring more accurate data processing for each channel.

[0046] The obtained RMS current value, current variation scale, heat dissipation efficiency fluctuation amplitude, average temperature rise rate, scene duration, and median switching frequency are stored in a structured format, i.e., in a database in an easily accessible and manageable format. Typically, these features are stored as a table or data structure, where each column corresponds to a feature and each row represents an operating scene or data point. This method facilitates subsequent querying, analysis, and processing of the stored feature data. The stored features are then integrated into the first operating condition feature, which represents the comprehensive performance of the equipment under all relevant features within the first operating scene slice.

[0047] Furthermore, by combining the multiple standard test scenarios with correlated perturbations, an intermediate test scenario array is obtained. The method includes:

[0048] Based on the device ID, an exhaustive search is performed to set test scenario switching, resulting in Q test scenario switching paths. Each test scenario switching path is a scene migration sequence from the starting scene to the target scene. The multiple standard test scenarios are mapped and combined according to the Q test scenario switching paths to obtain Q test scenario sequences. After perturbing the scene nodes of the Q test scenario sequences, intermediate state points are smoothly inserted to generate multiple intermediate scene sequences. The multiple intermediate scene sequences are aggregated to output the intermediate test scenario array.

[0049] Based on the device ID, an exhaustive search of test scenario switching settings is performed, listing all possible test scenario switching paths. This involves trying various combinations of test scenarios to explore the device's behavior under different operating conditions. A test scenario switching path refers to the process by which the device switches from one scenario to another. Each test scenario switching path includes multiple scenario migration steps, describing how the device transitions from one state to another under changing environmental conditions. Through the exhaustive search of test scenario switching settings, Q test scenario switching paths are obtained. Each path represents a migration process from the initial scenario to the target scenario. These paths encompass the device's behavioral trajectory under different loads, temperatures, and other operating conditions. The goal is to cover all possible states of the device under various operating conditions, thereby providing more comprehensive test conditions for subsequent dynamic wear testing.

[0050] The Q test scenario switching paths are mapped and combined with multiple standard test scenarios. Each test scenario switching path is combined with a standard test scenario to generate a corresponding complete test scenario sequence. In the test scenario sequence, multiple standard test scenarios are arranged in a certain order, reflecting the performance of the equipment under different operating conditions. Through the mapping process, Q test scenario sequences are obtained, and each sequence represents the complete working process of the equipment under different test conditions.

[0051] Cross-perturbation is performed on the scenario nodes in the Q test scenario sequences. A scenario node refers to each independent scenario in the test scenario sequence. Through cross-perturbation, the nodes are exchanged, transformed, or perturbed between different scenario nodes to simulate more complex or extreme working environments. The purpose is to create different working conditions and reflect the nonlinear behavior or abnormal situations that the equipment may encounter. The perturbed nodes help test the performance of the equipment in more varied scenarios and enhance the comprehensiveness of the test.

[0052] Between scene nodes after cross-perturbation, intermediate state points are smoothly inserted. This means inserting new intermediate scenes between two scene nodes to make the transition when the device switches from one scene to another smoother. Intermediate state points are used to simulate the transition states that the device may experience during scene switching. This smooth insertion can reduce the instability of device state changes and simulate more realistic working conditions. Through cross-perturbation and intermediate state point insertion, multiple intermediate scene sequences are generated. These sequences contain the process of smooth transition of the device between different working scenes and are a comprehensive representation of the device's state changes under various environments.

[0053] The generated intermediate scenario sequences are aggregated to integrate them into a larger test set, thereby ensuring coverage of more operating states and transition processes. After aggregation, an intermediate test scenario array is output, which contains multiple scenarios that have been perturbed, inserted, and transitioned, comprehensively covering the behavior of the device under various operating conditions.

[0054] Furthermore, after performing scene node cross-perturbation on the Q test scene sequences, intermediate state points are smoothly inserted to generate multiple intermediate scene sequences. The method includes:

[0055] The Q test scenario sequences are subjected to scenario node cross-perturbation simulation to generate Q sets of updated scenario sequences; physical feasibility verification screening is performed on the Q sets of updated scenario sequences to output a set of filtered scenario sequences, wherein the set of filtered scenario sequences includes the Q test scenario sequences; piecewise linear smooth interpolation is performed between adjacent nodes of the set of filtered scenario sequences to output the multiple intermediate scenario sequences.

[0056] Cross-perturbation simulation is performed on the scenario nodes in the Q test scenario sequences. Cross-perturbation simulation explores the possible performance of the device under different operating conditions by changing or alternating the scenario nodes in the test scenarios. The goal is to increase the diversity of scenarios, especially when testing the extreme states and boundary conditions of the device. Through cross-perturbation simulation, some new possible scenarios are generated. These scenarios represent the performance of the device under atypical operating conditions. After cross-perturbation simulation, Q sets of updated scenario sequences are generated. Compared with the original test scenario sequences, these updated scenario sequences have changes in some scenario nodes or state transitions, providing more test paths for subsequent tests.

[0057] Physical feasibility verification is performed on the Q-group update scenario sequences. This means checking whether each update scenario sequence meets the physical limitations and constraints of the actual equipment, such as whether physical quantities like voltage, current, and temperature are within reasonable ranges. Physical feasibility verification aims to eliminate unreasonable test scenarios or those exceeding the equipment's operating range, ensuring that only practically executable test paths are retained. For example, scenarios involving current overload or excessively high temperatures are filtered out. After physical feasibility verification, a set of filtered scenario sequences is output, containing all verified test scenario sequences. These sequences are not only theoretically reasonable but also conform to the physical characteristics of the actual equipment under different operating conditions.

[0058] Piecewise linear smooth interpolation is performed on each pair of adjacent nodes in the selected scene sequence set. This means inserting a smooth transition between two adjacent scene nodes, ensuring a more natural and fluid transition between different scenes. Piecewise linear interpolation creates a straight line between two data points to smooth the transition. This method helps avoid abrupt jumps when the device switches scenes, simulating more natural behavior changes of the device in real-world situations. Through interpolation, multiple intermediate scene sequences are generated. These intermediate scene sequences represent the smooth transition of the device from one operating state to another, and are particularly suitable for simulating the process of the device gradually adapting or changing in a dynamic environment.

[0059] Furthermore, after performing response delay tests on the CoolMOS devices in the device under test according to the multiple standard test scenarios, the intermediate state test scenario array is locally activated based on the test results. Dynamic loss tests on the CoolMOS devices are then performed on the device under test, and intermediate state multivariate loss trajectories are output. The method includes:

[0060] After scenario debugging using the aforementioned standard test scenarios, response delay tests are performed on the CoolMOS device in the device under test, yielding multiple delay test results. Multiple delay accommodating spaces within the aforementioned standard test scenarios are used to compare and determine the pass / fail status of the multiple delay test results, outputting N unqualified test scenarios. These N unqualified test scenarios are used as activation taboo anchor points to activate a local test subarray in the intermediate state test scenario array. The local test subarray is then divided into multiple short-thread scenario sequences. These short-thread scenario sequences are used to perform dynamic loss tests on the CoolMOS device in the device under test, outputting multiple short-thread loss trajectories, wherein these multiple short-thread loss trajectories constitute the intermediate state multi-dimensional loss trajectory.

[0061] Multiple standard test scenarios were used for scenario debugging to ensure the correct test environment and equipment configuration, and to ensure that the device's response under different test scenarios met expectations. For each standard test scenario, response delay testing was performed on the CoolMOS device in the device under test. Response delay testing measures the response time of the CoolMOS device after a change in the input signal. During the test, the time delay between receiving a change in the input signal and the change in the output signal was recorded. The device's response delay may vary due to the operating environment (such as temperature, load changes, etc.), therefore, testing under multiple standard test scenarios is necessary. Multiple delay test results were obtained, reflecting the device's delay performance under different operating conditions.

[0062] The system compares and judges multiple delay tolerance ranges across various standard test scenarios. Delay tolerance range refers to the maximum allowable error range or tolerance range for a device's response delay. During design, each standard test scenario pre-defines an allowable delay range, which determines the acceptable response delay for the device under that scenario. The delay test results are compared with the delay tolerance range corresponding to each standard test scenario. If a delay test result exceeds the tolerance range of that test scenario, the scenario is deemed unqualified. Finally, N unqualified test scenarios are output. These scenarios indicate that the response delay of the CoolMOS device under certain operating conditions exceeds the device's tolerance range, potentially indicating performance issues or areas requiring optimization.

[0063] The N unqualified test scenarios output are used as activation taboo anchors. Activation taboo anchors refer to critical scenarios that cannot obtain qualified results through the normal test path. These scenarios may reveal potential problems of the device under specific conditions. By using unqualified test scenarios as activation taboo anchors, we can focus on these special, poorly performing scenarios and further optimize the device's performance.

[0064] Based on the activation taboo anchor point, a local test subarray is activated within the intermediate-state test scenario array. The local test subarray refers to a subset of scenarios selected from the overall intermediate-state test scenario array, focusing on those failing test scenarios and their vicinity. The purpose of activating the local test subarray is to conduct more in-depth testing and analysis of these potentially problematic scenarios, evaluate the dynamic performance of CoolMOS devices under these specific scenarios, identify the causes, and optimize them.

[0065] The activated local test subarray is further divided into multiple short-thread scenario sequences. A local test subarray typically contains multiple scenarios focused on specific operating conditions. This division allows for more granular testing and analysis of these scenarios. A short-thread scenario sequence refers to a short test path, each containing a series of closely linked scenarios. This division allows for a more effective focus on the device's dynamic performance, especially its rapid response and energy efficiency under certain specific scenarios. Each short-thread scenario sequence represents the transition of the device from one operating state to another, used to evaluate the device's response and energy loss characteristics over a short period. By dividing the sequence, it is possible to focus on the device's short-term behavior under specific operating conditions within a localized scope, improving the accuracy of dynamic energy loss testing.

[0066] Dynamic loss testing of CoolMOS devices was conducted on the tested equipment across multiple short-thread scenario sequences. Each short-thread scenario sequence represents a short-term operating trajectory of the device under specific conditions. These trajectories were used to measure the loss characteristics of the CoolMOS devices. The dynamic loss test included measuring power loss, heat loss, and switching loss under various scenarios. Each short-thread scenario sequence corresponded to a specific test phase, aiming to reveal the loss behavior of the CoolMOS devices during transition periods. During the testing of each short-thread scenario sequence, various loss data were recorded, and the corresponding short-thread loss trajectory was output. The short-thread loss trajectory demonstrates the dynamic loss performance of the device within each short time period. By analyzing these short-thread loss trajectories, a better understanding of the device's energy efficiency under different operating states can be achieved. Ultimately, all short-thread loss trajectories formed a complete intermediate-state multi-dimensional loss trajectory. This intermediate-state multi-dimensional loss trajectory encompasses all loss data of the device in different short-term transition scenarios, providing a comprehensive energy efficiency assessment of the device under multiple dynamic operating conditions.

[0067] Furthermore, the method involves using the multiple short-thread scenario sequences to perform dynamic loss testing on the CoolMOS device in the device under test, and outputting multiple short-thread loss trajectories.

[0068] During the dynamic loss test of the CoolMOS device on the device under test, a first short-thread scenario sequence is used to synchronously acquire a first electrical waveform sequence and a first thermal distribution sequence; loss is solved based on the first electrical waveform sequence and the first thermal distribution sequence to output the first conduction power loss and the first switching energy loss; the first conduction power loss and the first switching energy loss are timestamped and fused to output the first short-thread loss trajectory.

[0069] The first short-thread scenario sequence is any one of multiple short-thread scenario sequences and is used as the current analysis object. The first short-thread scenario sequence is used to perform dynamic loss testing on the CoolMOS device in the device under test. During the test, the first electrical waveform sequence and the first thermal distribution sequence are acquired simultaneously. The first electrical waveform sequence records the changes in electrical characteristics of the device during operation, such as voltage and current. The electrical waveform is used to analyze the power loss, switching response, and other characteristics of the CoolMOS device under different operating states. The first thermal distribution sequence records the temperature changes of the device during the dynamic test, especially the temperature rise of the CoolMOS device. The temperature distribution sequence is used to evaluate the magnitude of heat loss and thermal management efficiency of the device during load changes.

[0070] The loss calculation is based on the first electrical waveform sequence and the first thermal distribution sequence. Specifically, the first electrical waveform sequence is used to calculate the conduction power loss and switching energy loss. The conduction power loss refers to the power loss of the CoolMOS device when current flows through its conductor in the conduction state; the switching energy loss refers to the energy loss generated by the device during the switching process. The first thermal distribution sequence is used to assist in the loss calculation, especially the relationship between temperature change and electrical waveform. By correlating temperature and current, the heat loss caused by current and switching operation can be estimated more accurately. After the loss calculation, the first conduction power loss and the first switching energy loss are output. The first conduction power loss is the power loss generated by the CoolMOS device during conduction due to current flowing through the device, which is usually related to the magnitude of the current and the on-resistance of the device; the first switching energy loss is the energy loss consumed by the CoolMOS device during switching, which is related to factors such as switching frequency and switching time.

[0071] The first data sequences of conduction power loss and first switching energy loss are timestamped. Since these two loss characteristics may deviate in time—for example, changes in current waveform and heat distribution may not be perfectly synchronized—timestamp alignment is necessary to ensure data accuracy and consistency at the same point in time. The aligned data is then fused; that is, the time sequences of conduction power loss and switching energy loss are combined into a complete loss data trajectory. This fusion is used to comprehensively evaluate the overall loss of the CoolMOS device under this test scenario. Finally, the first short-thread loss trajectory is output, which is a complete trajectory combining conduction power loss and switching energy loss.

[0072] Furthermore, the method involves performing spatiotemporal evolution modeling on the intermediate-state multivariate loss trajectory and outputting a dynamic loss cloud map, the method comprising:

[0073] Construct a spatiotemporal evolution tensor; perform loss evolution fusion modeling on the intermediate state multivariate loss trajectory based on the spatiotemporal evolution tensor, and output the dynamic loss cloud map.

[0074] The spatiotemporal evolution tensor is a mathematical structure used to represent data that changes over time and space (or device state, etc.). It can capture dynamic behavior in multiple dimensions and is suitable for complex data analysis involving changes in time and space.

[0075] By utilizing the spatiotemporal evolution tensor to perform loss evolution fusion modeling on previously obtained intermediate-state multivariate loss trajectories, the aim is to reveal the loss evolution process of CoolMOS devices under different operating conditions. Loss evolution fusion modeling, by fusing loss data across time and space dimensions, provides a more comprehensive understanding of the changing trends of loss with factors such as time, load, and temperature. This modeling approach not only helps analyze the total change in loss but also reveals loss fluctuations and trends under different operating conditions. The dynamic loss cloud map is the output of the loss evolution modeling; it is a visual representation of the device's loss process. The results generated through loss evolution fusion modeling can present loss data graphically, using heatmaps or 3D graphics to represent the loss changes of the device under different operating conditions. This dynamic loss cloud map displays the loss distribution of CoolMOS devices under different operating conditions, helping analysts intuitively see which operating conditions or time periods have higher losses and which operating conditions show more efficient device performance.

[0076] In summary, the dynamic loss detection method for CoolMOS devices provided in this application has the following technical effects:

[0077] By retrieving and aggregating historical usage scenarios, multiple standard test scenarios are obtained. These scenarios represent various working environments that the device may face in actual use. The aggregation process ensures the comprehensiveness and diversity of the test scenarios, enabling the tests to cover the device's performance under different conditions. By combining these standard test scenarios with correlated perturbations, an intermediate test scenario array is generated. These perturbation combinations simulate the dynamic changes of the device under various operating conditions, further increasing the diversity and complexity of the tests, thereby improving their realism and accuracy. By conducting response delay tests on the CoolMOS devices in the device under test based on these standard test scenarios, the response time of the CoolMOS devices under different operating conditions can be evaluated, providing the device with... Based on the test results, specific intermediate-state test scenario arrays are locally activated to assess latency performance under different loads. This local activation allows for refined testing of potential performance bottlenecks or critical scenarios, further outputting intermediate-state multi-dimensional loss trajectories. This makes the dynamic loss assessment more precise and realistic, enabling the identification of potential performance issues or optimization opportunities under specific conditions. By modeling the spatiotemporal evolution of the intermediate-state multi-dimensional loss trajectories, dynamic loss cloud maps can be generated by combining temporal and spatial changes. These dynamic loss cloud maps can intuitively display the loss evolution trend of the device under different operating conditions, helping engineers quickly identify high-loss regions of the device under different times and operating conditions, guiding subsequent optimization work, and thus effectively improving the energy efficiency and stability of CoolMOS devices.

[0078] Example 2, based on the same inventive concept as the dynamic loss detection method for CoolMOS devices in the foregoing examples, such as... Figure 2 As shown in the figure, this application provides a dynamic loss detection system for CoolMOS devices, the system comprising:

[0079] The standard test scenario acquisition module 10 is used to retrieve historical usage scenarios based on the device ID and obtain multiple standard test scenarios by aggregating the historical usage scenarios.

[0080] The correlation perturbation combination module 20 is used to obtain an intermediate test scenario array by performing correlation perturbation combination on the multiple standard test scenarios.

[0081] The dynamic loss test module 30 is used to perform dynamic loss test on the CoolMOS device in the device under test after performing response delay test according to the multiple standard test scenarios, and to locally activate the intermediate state test scenario array according to the test results, and to perform dynamic loss test on the CoolMOS device in the device under test, and output the intermediate state multi-dimensional loss trajectory.

[0082] The spatiotemporal evolution modeling module 40 is used to perform spatiotemporal evolution modeling on the intermediate state multivariate loss trajectory and output a dynamic loss cloud map.

[0083] Furthermore, the standard test scenario acquisition module 10 is used to perform the following operation steps:

[0084] Based on the device ID, operation logs of devices of the same model are collected to obtain a feature device log set; the feature device log set is segmented into scenarios based on physical event triggering to obtain multiple operation scenario slices; after extracting the operating condition feature vectors from the multiple operation scenario slices, operating condition aggregation is performed based on the extraction results to output the multiple standard test scenarios.

[0085] Furthermore, the standard test scenario acquisition module 10 is used to perform the following operation steps:

[0086] A pre-constructed operating condition feature extraction model is used, wherein the operating condition feature extraction model has built-in parallel isolated electrical feature extraction channels, thermal dynamic feature extraction channels, and time feature extraction channels; a first operating scenario slice is loaded into the operating condition feature extraction model, and the first operating condition feature is analyzed and output, wherein the first operating condition feature consists of the effective value of current, the scale of current change, the amplitude of heat dissipation efficiency fluctuation, the average temperature rise rate, the duration of the scenario, and the median switching frequency; multiple operating scenario slices are input into the operating condition feature extraction model, and operating condition feature vectors are extracted by analogy to obtain multiple operating condition features; operating condition clustering based on recurrence density is performed on the multiple operating condition features to output the multiple standard test scenarios.

[0087] Furthermore, the standard test scenario acquisition module 10 is used to perform the following operation steps:

[0088] After loading the first operating scenario slice into the operating condition feature extraction model: Step a: Extract the average test voltage through the voltage feature sampling unit in the electrical feature extraction channel, and extract the effective value of the current and the current change scale through the current feature sampling unit; Step b: Calculate the heat dissipation efficiency fluctuation amplitude and the average temperature rise rate through the thermal dynamic feature extraction channel; Step c: Calculate the scene duration and the median switching frequency through the time feature extraction channel; wherein, steps a, b, and c are executed in parallel and in isolation; the first operating condition features are obtained by structured storage of the effective value of the current, the current change scale, the heat dissipation efficiency fluctuation amplitude, the average temperature rise rate, the scene duration, and the median switching frequency.

[0089] Furthermore, the correlation perturbation combination module 20 is used to perform the following operation steps:

[0090] Based on the device ID, an exhaustive search is performed to set test scenario switching, resulting in Q test scenario switching paths. Each test scenario switching path is a scene migration sequence from the starting scene to the target scene. The multiple standard test scenarios are mapped and combined according to the Q test scenario switching paths to obtain Q test scenario sequences. After perturbing the scene nodes of the Q test scenario sequences, intermediate state points are smoothly inserted to generate multiple intermediate scene sequences. The multiple intermediate scene sequences are aggregated to output the intermediate test scenario array.

[0091] Furthermore, the correlation perturbation combination module 20 is used to perform the following operation steps:

[0092] The Q test scenario sequences are subjected to scenario node cross-perturbation simulation to generate Q sets of updated scenario sequences; physical feasibility verification screening is performed on the Q sets of updated scenario sequences to output a set of filtered scenario sequences, wherein the set of filtered scenario sequences includes the Q test scenario sequences; piecewise linear smooth interpolation is performed between adjacent nodes of the set of filtered scenario sequences to output the multiple intermediate scenario sequences.

[0093] Furthermore, the dynamic loss test module 30 is used to perform the following operation steps:

[0094] After scenario debugging using the aforementioned standard test scenarios, response delay tests are performed on the CoolMOS device in the device under test, yielding multiple delay test results. Multiple delay accommodating spaces within the aforementioned standard test scenarios are used to compare and determine the pass / fail status of the multiple delay test results, outputting N unqualified test scenarios. These N unqualified test scenarios are used as activation taboo anchor points to activate a local test subarray in the intermediate state test scenario array. The local test subarray is then divided into multiple short-thread scenario sequences. These short-thread scenario sequences are used to perform dynamic loss tests on the CoolMOS device in the device under test, outputting multiple short-thread loss trajectories, wherein these multiple short-thread loss trajectories constitute the intermediate state multi-dimensional loss trajectory.

[0095] Furthermore, the dynamic loss test module 30 is used to perform the following operation steps:

[0096] During the dynamic loss test of the CoolMOS device on the device under test, a first short-thread scenario sequence is used to synchronously acquire a first electrical waveform sequence and a first thermal distribution sequence; loss is solved based on the first electrical waveform sequence and the first thermal distribution sequence to output the first conduction power loss and the first switching energy loss; the first conduction power loss and the first switching energy loss are timestamped and fused to output the first short-thread loss trajectory.

[0097] Furthermore, the spatiotemporal evolution modeling module 40 is used to perform the following operation steps:

[0098] Construct a spatiotemporal evolution tensor; perform loss evolution fusion modeling on the intermediate state multivariate loss trajectory based on the spatiotemporal evolution tensor, and output the dynamic loss cloud map.

[0099] Through the foregoing detailed description of the dynamic loss detection method for CoolMOS devices, those skilled in the art can clearly understand the dynamic loss detection system for CoolMOS devices in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0100] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for dynamic loss detection for CoolMOS devices, characterized in that, The method comprises: According to the on-board equipment ID, the historical use scene is retrieved, and a plurality of standard test scenes are obtained by aggregating the historical use scene; By associating and disturbing the plurality of standard test scenes, an intermediate state test scene array is obtained; After the response delay test of the CoolMOS device in the to-be-tested on-board equipment according to the plurality of standard test scenes is performed, the intermediate state test scene array is activated locally according to the test result, and the dynamic loss test of the CoolMOS device in the to-be-tested on-board equipment is performed, to output an intermediate state multi-element loss track; The intermediate state multi-element loss track is subjected to time-space evolution modeling to output a dynamic loss cloud chart; By associating and disturbing the plurality of standard test scenes, an intermediate state test scene array is obtained, and the method comprises: According to the on-board equipment ID, a test scene switching setting exhaustion is performed to obtain Q test scene switching paths, wherein each test scene switching path is a scene migration sequence from a starting scene to a target scene; The plurality of standard test scenes are mapped and combined according to the Q test scene switching paths to obtain Q test scene sequences; After the Q test scene sequences are subjected to scene node cross disturbance, intermediate state point smoothing insertion is performed to generate a plurality of intermediate state scene sequences; The plurality of intermediate state scene sequences are aggregated to output the intermediate state test scene array; After the response delay test of the CoolMOS device in the to-be-tested on-board equipment according to the plurality of standard test scenes is performed, the intermediate state test scene array is activated locally according to the test result, and the dynamic loss test of the CoolMOS device in the to-be-tested on-board equipment is performed, to output an intermediate state multi-element loss track, the method comprising: After the scene debugging is performed by using the plurality of standard test scenes, the response delay test of the CoolMOS device in the to-be-tested on-board equipment is performed to obtain a plurality of delay test results; The plurality of delay test results are subjected to qualifiedness comparison and judgment by calling a plurality of delay accommodation spaces of the plurality of standard test scenes, to output N unqualified test scenes; The N unqualified test scenes are taken as activation taboo anchor points to activate a local test subarray in the intermediate state test scene array, and the local test subarray focuses on unqualified test scenes and scenes near the unqualified test scenes; The local test subarray is split into a plurality of short thread scene sequences; The dynamic loss test of the CoolMOS device in the to-be-tested on-board equipment is performed by using the plurality of short thread scene sequences to output a plurality of short thread loss tracks, wherein the plurality of short thread loss tracks constitute the intermediate state multi-element loss track.

2. A method for dynamic loss detection for a CoolMOS device as defined in claim 1, wherein, According to the on-board equipment ID, the historical use scene is retrieved, and a plurality of standard test scenes are obtained by aggregating the historical use scene, the method comprising: According to the on-board equipment ID, the operation log collection of the same type of equipment is performed to obtain a feature equipment log set; The feature equipment log set is subjected to scene cutting based on physical event triggering to obtain a plurality of operation scene slices; After the working condition characteristic vector extraction is performed on the plurality of running scene slices, working condition aggregation is performed according to the extraction result, and the plurality of standard test scenes are output.

3. A method for dynamic loss detection for a CoolMOS device as defined in claim 2, wherein, After the working condition characteristic vector extraction is performed on the plurality of running scene slices, working condition aggregation is performed according to the extraction result, and the plurality of standard test scenes are output, the method comprises: A pre-constructed working condition characteristic extraction model, wherein the working condition characteristic extraction model is built-in and parallel-isolated electrical characteristic extraction channels, thermal dynamic characteristic extraction channels and time characteristic extraction channels; Load the first running scene slice into the working condition characteristic extraction model, and analyze and output the first running working condition characteristic, wherein the first running working condition characteristic is composed of the current effective value, the current change scale, the heat dissipation efficiency fluctuation amplitude, the temperature rise rate average value, the scene duration and the switching frequency median value; The plurality of running scene slices are input into the working condition characteristic extraction model, and working condition characteristic vector extraction is performed by analogy to obtain a plurality of running working condition characteristics; Perform working condition clustering based on the recurrence density on the plurality of running working condition characteristics, and output the plurality of standard test scenes.

4. The method for dynamic loss detection of a CoolMOS device of claim 3, wherein, Load the first running scene slice into the working condition characteristic extraction model, and analyze and output the first running working condition characteristic, the method comprising: After loading the first running scene slice into the working condition characteristic extraction model: Step a: extracting the test voltage average value through the voltage characteristic sampling unit in the electrical characteristic extraction channel, extracting the current effective value and the current change scale through the current characteristic sampling unit; Step b: calculating the heat dissipation efficiency fluctuation amplitude and the temperature rise rate average value through the thermal dynamic characteristic extraction channel; Step c: calculating the scene duration and the switching frequency median value through the time characteristic extraction channel; Wherein, step a, step b and step c are executed in parallel isolation; Obtaining the first running working condition characteristic by structurally storing the current effective value, the current change scale, the heat dissipation efficiency fluctuation amplitude, the temperature rise rate average value, the scene duration and the switching frequency median value.

5. The method for dynamic loss detection of a CoolMOS device of claim 1, wherein, After the scene node cross disturbance is performed on the Q test scene sequences, intermediate state point smoothing insertion is performed to generate a plurality of intermediate state scene sequences, the method comprising: Performing scene node cross disturbance simulation on the Q test scene sequences to expand and generate Q groups of updated scene sequences; Performing physical feasibility verification screening on the Q groups of updated scene sequences to output a screening scene sequence set, wherein the screening scene sequence set comprises the Q test scene sequences; Performing segmented linear smoothing interpolation between adjacent nodes of the sequence to output the plurality of intermediate state scene sequences.

6. The method for dynamic loss detection of a CoolMOS device of claim 1, wherein, Using the plurality of short thread scene sequences to perform dynamic loss testing of CoolMOS devices on the to-be-detected mounted device, outputting a plurality of short thread loss trajectories, the method comprising: During the dynamic loss testing of CoolMOS devices on the to-be-detected mounted device using the first short thread scene sequence, synchronously collecting a first electrical waveform sequence and a first thermal distribution sequence; Based on the first electrical waveform sequence and the first thermal distribution sequence, loss is solved to output the first conduction power loss and the first switching energy loss; The first conduction power loss and the first switch energy loss are time-stamped aligned and fused, and a first short thread loss trajectory is output.

7. The method for dynamic loss detection of a CoolMOS device of claim 6, wherein, The intermediate state multi-element loss trajectory is modeled for space-time evolution, and a dynamic loss cloud map is output, the method comprising: constructing a space-time evolution tensor; based on the space-time evolution tensor, the intermediate state multi-element loss trajectory is modeled for loss evolution fusion, and the dynamic loss cloud map is output.

8. A dynamic loss detection system for a CoolMOS device, characterized by, The system for implementing the dynamic loss detection method for CoolMOS devices according to any one of claims 1-7, comprising: a standard test scenario acquisition module for retrieving historical use scenarios according to the on-board device ID, and obtaining a plurality of standard test scenarios by aggregating the historical use scenarios; an association disturbance combination module for obtaining an intermediate state test scenario array by associating and disturbing the plurality of standard test scenarios; a dynamic loss test module for locally activating the intermediate state test scenario array according to the test results after the response delay test of the CoolMOS device in the to-be-detected on-board device according to the plurality of standard test scenarios, and performing dynamic loss test of the CoolMOS device in the to-be-detected on-board device, and outputting an intermediate state multi-element loss trajectory; a space-time evolution modeling module for modeling the intermediate state multi-element loss trajectory for space-time evolution, and outputting a dynamic loss cloud map.

Images