A fast testing method and system for MOSFET devices

Abstract

The present invention relates to the field of transistor detection technology, and in particular to a fast testing method and system for MOSFET devices. The method comprises the following steps: obtaining and integrating multi-source parallel data of MOSFET devices to obtain real-time device I-V characteristic data; predicting MOSFET device performance based on real-time device I-V characteristic data to obtain MOSFET device performance prediction data; screening according to MOSFET device performance prediction data to obtain preliminary screening qualified device performance data; adjusting based on preliminary screening qualified device performance data to obtain MOSFET device test plan; obtaining real-time MOSFET device test data, and screening qualified devices to obtain re-screening qualified device performance data; summarizing device tests based on preliminary screening qualified device performance data and re-screening qualified device performance data to obtain MOSFET device test report. The present invention can shorten the test cycle.

Description

Technical Field

[0001] The present invention relates to the technical field of transistor detection, and in particular to a fast testing method and system for MOSFET devices. Background Art

[0002] Semiconductor devices, especially metal oxide semiconductor field effect transistors (MOSFETs), are widely and far-reachingly used in modern electronic devices. With the rapid development of integrated circuit technology, the manufacture and testing of MOSFETs have become an important part of electronic product production and quality control. In the traditional MOSFET test process, parameter test instruments are usually used to measure the main performance indicators of the device, including threshold voltage (V_th), transduction (gm), leakage current (Idss) and other parameters. The test method generally relies on current-voltage (IV) characteristic measurement. During the measurement process, the source, drain, gate and substrate (for bipolar MOSFET) of the MOSFET need to be connected to different test signals. By applying a specific combination of gate voltage and source-drain voltage, the current response of the device is obtained, and then the working performance of the device is evaluated. This traditional test method can comprehensively and accurately evaluate the static and dynamic characteristics of MOSFETs with high accuracy and reliability. Traditional methods usually require multi-point measurements to obtain IV characteristic curves. The whole process involves multiple voltage steps, resulting in a long test time. Especially in large-scale production processes, testing each device one by one will significantly increase the production cycle and reduce production efficiency. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide a fast testing method and system for MOSFET devices to solve at least one of the above technical problems.

[0004] To achieve the above object, a fast testing method for a MOSFET device comprises the following steps:

[0005] Step S1: acquiring multi-source parallel data of MOSFET devices, and performing data preprocessing according to the multi-source parallel data of MOSFET devices, thereby obtaining multi-source parallel data to be analyzed; integrating device IV characteristics of the multi-source parallel data to be analyzed, thereby obtaining real-time device IV characteristic data;

[0006] Step S2: performing a voltage-current response analysis based on the real-time device IV characteristic data to obtain a real-time device voltage-current response curve; performing a MOSFET device performance prediction on the real-time device voltage-current response curve to obtain MOSFET device performance prediction data;

[0007] Step S3: Screen qualified devices according to the MOSFET device performance prediction data, so as to obtain the performance data of the preliminarily screened qualified devices; adaptively adjust the test application parameters based on the preliminarily screened qualified device performance data, so as to obtain the MOSFET device test plan, and upload it to the MOSFET device test management platform to execute the test plan implementation task;

[0008] Step S4: acquiring real-time test data of MOSFET devices, and performing device performance evaluation based on the real-time test data of MOSFET devices, thereby obtaining device performance evaluation data; performing qualified device screening according to the device performance evaluation data, thereby obtaining re-screened qualified device performance data;

[0009] Step S5: Summarize the device tests based on the performance data of the devices that have passed the initial screening and the performance data of the devices that have passed the re-screening, so as to obtain a MOSFET device test report.

[0010] The present invention can improve the integration and analysis efficiency of data by acquiring multi-source parallel data of MOSFET devices and performing data preprocessing, so that test information from different sources can be effectively summarized, thereby avoiding the inefficiency of scattered test data processing in traditional methods. Real-time acquisition of IV characteristic data can accelerate the real-time performance of the test process, reflect device performance in real time, enable testing and data processing to be carried out synchronously, and reduce the delay of waiting and processing time. By performing voltage-current response analysis and device performance prediction based on real-time IV characteristic data, not only can the working characteristics of the device be accurately identified, but also the performance trend of the device can be provided in the prediction stage, which is helpful to identify potential quality problems in advance and ensure the efficiency of product quality control. The acquisition of performance prediction data is helpful to adjust the test plan in advance to cope with different device characteristics, and improves the accuracy and pertinence of the test. Screening qualified devices based on performance prediction data can effectively reduce the test burden of unqualified devices, optimize the configuration of test resources, avoid invalid repeated tests, and thus improve production efficiency. The adaptive adjustment of the test plan further improves the personalization and accuracy of the test, so that each device can accept a tailored test plan according to its characteristics, which is often difficult to achieve in traditional manual adjustment. During the real-time testing process, performance evaluation and re-screening based on real-time data can further improve the accuracy and reliability of the test and reduce misjudgments caused by test errors or external environmental factors. The re-screening process ensures that the performance data of the final qualified device is more in line with product requirements and improves the consistency of device quality. The final test summary report not only provides comprehensive data support for product quality, but also provides a reliable basis for subsequent production adjustments and optimizations, and can achieve data-based refined production management and quality control. Overall, this series of steps greatly shortens the test cycle, reduces production costs, and ensures the reliability and consistency of product quality through intelligent data processing, real-time testing and predictive analysis, providing strong technical support for the production of modern semiconductor devices.

[0011] Optionally, step S1 specifically includes:

[0012] Step S11: acquiring multi-source parallel data of MOSFET devices, and performing data preprocessing on the multi-source parallel data of MOSFET devices, thereby obtaining multi-source parallel data to be analyzed;

[0013] Step S12: performing linear data division on the multi-source parallel data to be analyzed, thereby obtaining linear multi-source parallel data and nonlinear multi-source parallel data;

[0014] Step S13: performing IV characteristic polynomial fitting on the linear multi-source parallel data to obtain linear IV characteristic data; performing IV characteristic exponential fitting on the nonlinear multi-source parallel data to obtain nonlinear IV characteristic data;

[0015] Step S14: merging multi-source data based on the linear IV characteristic data and the nonlinear IV characteristic data to obtain device IV characteristic data;

[0016] Step S15: performing real-time data integration on the multi-source parallel data to be analyzed according to the device IV characteristic data, thereby obtaining real-time device IV characteristic data.

[0017] The present invention can effectively integrate data from different sources by acquiring multi-source parallel data of MOSFET devices and performing data preprocessing, thereby ensuring the integrity and consistency of data, thereby providing high-quality input for subsequent analysis. When performing data division, the data is divided into linear and nonlinear parts, which is helpful to adopt appropriate fitting methods for different types of data and improve the accuracy of data analysis. The result obtained by polynomial fitting of linear IV characteristic data can provide a clear current-voltage relationship, which is suitable for the analysis of linear working areas, while the nonlinear IV characteristic data can more accurately describe the performance of the device in the nonlinear working area through exponential fitting, which makes up for the limitations of traditional linear fitting methods under complex working conditions. By merging linear and nonlinear IV characteristic data, more comprehensive device characteristic data can be obtained, which improves the accuracy and credibility of the analysis results. On this basis, through real-time data integration, the working state of the device can be dynamically updated and tracked, ensuring the accuracy of real-time analysis, and then providing strong support for the performance optimization and fault diagnosis of the device. The whole process effectively improves the comprehensiveness, accuracy and timeliness of the IV characteristic analysis of MOSFET devices through multi-level and multi-angle processing of data.

[0018] Optionally, step S14 is specifically:

[0019] Step S141: dividing the linear IV characteristic data and the nonlinear IV characteristic data into acquisition time windows respectively, thereby obtaining windowed linear IV characteristic data and windowed nonlinear IV characteristic data;

[0020] Step S142: performing window spectrum conversion on the windowed linear IV characteristic data and the windowed nonlinear IV characteristic data respectively, so as to obtain a linear IV characteristic spectrum and a nonlinear IV characteristic spectrum;

[0021] Step S143: Calculating characteristic spectrum similarity according to the linear IV characteristic spectrum and the nonlinear IV characteristic spectrum, thereby obtaining characteristic spectrum similarity data;

[0022] Step S144: identifying characteristic segment merging points according to the characteristic spectrum similarity data, thereby obtaining characteristic segment merging point data;

[0023] Step S145: performing characteristic segmentation merging on the linear IV characteristic data and the nonlinear IV characteristic data based on the characteristic segmentation merging point data, thereby obtaining device IV characteristic data.

[0024] The present invention can realize local analysis of data by dividing linear and nonlinear IV characteristic data into time windows, so that data characteristics of different time periods can be more accurately identified. This division makes data analysis more flexible, and data can be processed according to different time periods, thereby avoiding errors that may be caused by global processing. This step can classify linear and nonlinear IV characteristics into their respective analysis frameworks, which is convenient for subsequent spectrum conversion and similarity calculation. Through spectrum conversion, frequency domain analysis of linear and nonlinear IV characteristics in the window can more deeply reveal the periodicity and frequency characteristics in the data. Spectral analysis can understand IV characteristics from another perspective, especially can distinguish small changes or implicit nonlinear behaviors therein, and provide more reliable frequency information for subsequent similarity calculation and characteristic merging. Similarity calculation based on spectrum can quantify the similarity between linear and nonlinear IV characteristics. This calculation helps to extract similar patterns from different characteristic data, and can effectively identify commonalities and differences in data. This provides a theoretical basis for characteristic segmentation merging, helps to efficiently classify similar characteristics into one segment, and avoids the inconsistency caused by artificial division. In the process of identifying the characteristic segment merging points, by accurately calculating the similarity, the key change points in the linear and nonlinear IV characteristic data can be effectively found, so that different characteristic regions can be accurately merged. This not only reduces redundant data processing, but also improves the processing efficiency and accuracy of the data, which is helpful for the subsequent unified description and modeling of the device IV characteristics. Through the process of characteristic segment merging, the linear and nonlinear IV characteristic data can be combined to form a complete device IV characteristic description. This integration process is of great significance to improving the accuracy of device performance analysis and the intelligence of data processing, and it can also provide a solid data foundation for more complex device characteristic modeling and optimization.

[0025] Optionally, step S2 specifically includes:

[0026] Step S21: constructing a voltage-current characteristic model based on real-time device IV characteristic data, thereby obtaining a voltage-current characteristic model;

[0027] Step S22: performing MOSFET device test mode recognition on the MOSFET device multi-source parallel data, thereby obtaining MOSFET device test mode data;

[0028] Step S23: performing a pole voltage mode simulation on the voltage-current characteristic model according to the MOSFET device test mode data, thereby obtaining pole voltage mode test simulation data;

[0029] Step S24: performing real-time device voltage-current response backtracking based on the pole voltage mode test simulation data, thereby obtaining a real-time device voltage-current response curve;

[0030] Step S25: performing MOSFET device performance prediction on the real-time device voltage-current response curve, thereby obtaining MOSFET device performance prediction data.

[0031] The present invention can realize accurate modeling of MOSFET devices by constructing a voltage-current characteristic model based on real-time device IV characteristic data, so that the electrical behavior of the device can be effectively simulated in subsequent tests and applications, and a reliable data basis is provided for performance evaluation under different working conditions. Test mode recognition is performed using multi-source parallel data of MOSFET devices, which can provide multi-dimensional test data support for different working modes, thereby improving the accuracy of model prediction. By simulating the voltage-current characteristic model in extreme voltage mode, the working performance of the device under extreme conditions can be deeply explored, and a key performance evaluation tool can be provided for designers to avoid failure or performance degradation caused by overload phenomena under extreme voltage conditions. The generation of extreme voltage mode test simulation data provides a strong basis for subsequent real-time device voltage-current response backtracking, so that the response characteristics in actual applications can be accurately predicted before the equipment is put into use, which provides technical guarantee for real-time monitoring and maintenance. By predicting the performance of MOSFET devices based on the real-time device voltage-current response curve, a comprehensive analysis and evaluation of device performance can be achieved, thereby effectively improving the reliability and life prediction of the device in actual applications, and avoiding system crashes or adverse effects caused by unstable performance or unexpected failures.

[0032] Optionally, step S24 is specifically:

[0033] Performing voltage-current response curve conversion according to the pole voltage mode test simulation data to obtain a voltage-current simulation response curve, and performing frequency domain conversion on the voltage-current simulation response curve to obtain a pole voltage test response spectrum;

[0034] Classify the frequency jitter characteristics according to the extreme voltage test response spectrum to obtain a smooth jitter spectrum and a sharp jitter spectrum;

[0035] Based on the stable jitter spectrum, the platform area curve of the voltage-current simulation response curve is identified to obtain the platform area response curve;

[0036] Based on the sharp jitter spectrum, the breakdown region curve is identified on the voltage-current simulation response curve, so as to obtain the breakdown region response curve;

[0037] Performing current peak identification on the voltage-current simulation response curve to obtain current peak data;

[0038] The pole voltage and current characteristics are integrated according to the platform area response curve, the collapse area response curve and the current peak data, so as to obtain the pole voltage and current characteristic data;

[0039] Through the voltage-current characteristic model, the real-time device voltage-current response is traced back to the real-time device IV characteristic data according to the pole voltage and current characteristic data, so as to obtain the real-time device voltage-current response curve.

[0040] The present invention can extract the response characteristics of the system at different frequencies by performing frequency domain conversion on the voltage-current response curve, laying the foundation for the subsequent frequency jitter feature classification. The application of frequency jitter feature classification makes it possible to identify and divide different working intervals of the voltage-current response curve according to the two spectrum characteristics of steady jitter and sharp jitter. The steady jitter spectrum helps to identify the stability area of ​​the system, that is, the platform area response curve. The characteristics of these areas can accurately judge the steady-state performance of the system when it is working. The sharp jitter spectrum can identify the behavior of the system entering the collapse area, which provides key information for the system's extreme performance, fault warning, etc. Through the current peak recognition technology, the current peak of the system during the voltage change process can be accurately extracted, thereby providing important data support for subsequent characteristic analysis. The integration of the response curves of the platform area and the collapse area and the current peak data enables us to evaluate the working performance of the system as a whole, especially the performance of the voltage and current characteristics under different states. The integrated pole voltage and current characteristic data can provide support for real-time backtracking based on the voltage-current characteristic model, so that during the real-time device operation, the voltage-current response curve of the device can be quickly traced back according to the current current and voltage characteristics, thereby helping to determine the working status of the device, predict possible failures, or optimize device performance.

[0041] Optionally, step S25 is specifically:

[0042] Performing breakdown region curve identification and platform region curve identification on the real-time device voltage-current response curve, thereby obtaining a real-time breakdown region curve and a real-time platform region curve;

[0043] Calculating the breakdown region curve ratio of the real-time device voltage-current response curve and the real-time breakdown region curve to obtain breakdown region curve ratio data;

[0044] Calculate the platform area curve ratio of the real-time device voltage-current response curve and the real-time platform area curve, so as to obtain platform area curve ratio data;

[0045] The device electrical performance is predicted based on the breakdown area curve proportion data and the platform area curve proportion data, thereby obtaining the MOSFET device performance prediction data.

[0046] The present invention can accurately identify the breakdown zone and platform zone of the device under different working conditions by analyzing the real-time device voltage-current response curve, thereby effectively obtaining real-time breakdown zone and platform zone curves. These curves can reflect the critical state and stable state that may occur in the device during operation, and provide key data for subsequent performance analysis. By calculating the proportion of the breakdown zone curve and the platform zone curve, the degree to which the device enters different states during operation is further quantified, which provides an effective quantitative indicator for the stability and reliability analysis of the device. In addition, using the breakdown zone and platform zone curve proportion data, the electrical performance of the device can be accurately predicted, thereby realizing the prediction and optimization of the performance of devices such as MOSFET in practical applications.

[0047] Optionally, step S3 specifically includes:

[0048] Step S31: Calculating the electrical performance index of the MOSFET device according to the MOSFET device performance prediction data, thereby obtaining the MOSFET device prediction electrical performance index data;

[0049] Step S32: Screening the MOSFET device predicted electrical performance index data for qualified devices based on the preset MOSFET device qualified electrical performance index, thereby obtaining the performance data of the preliminarily screened qualified devices and the performance data of the preliminarily screened unqualified devices;

[0050] Step S33: acquiring historical test data of the MOSFET device, and dividing the historical test data of the MOSFET device into test characteristics, thereby obtaining a historical test application parameter set and historical test result data;

[0051] Step S34: Calculate the Spearman correlation coefficient for the historical test application parameter set and the historical test result data, so as to obtain application parameter-test result correlation data;

[0052] Step S35: Adaptively adjust the test application parameters of the initially screened qualified device performance data according to the application parameter-test result correlation data, thereby obtaining a MOSFET device test plan, and upload it to the MOSFET device test management platform to execute the test plan implementation task.

[0053] The present invention provides scientific and systematic data support for the evaluation of device performance by calculating electrical performance indicators based on MOSFET device performance prediction data, making the preliminary prediction of device performance more accurate, avoiding over-reliance on traditional test methods, and thus improving development and evaluation efficiency. By performing qualified screening on the predicted electrical performance data according to preset qualified electrical performance indicators, qualified devices that meet the standards can be effectively screened out from a large number of devices to be tested, reducing the workload of subsequent testing, while reducing the waste and testing costs of unqualified devices, and improving production efficiency. This also helps to ensure the centralization and pertinence of subsequent tests, laying a foundation for further optimizing product quality. By acquiring and dividing historical test data, historical test experience can be accumulated, providing a large amount of real test data support for subsequent predictions, thereby enhancing the reliability of the test scheme. By rationally analyzing historical data and calculating the Spearman correlation coefficient, the potential relationship between test parameters and test results can be revealed, which provides a scientific basis for further optimizing the test scheme and reduces the errors and uncertainties that may exist in the test. By adaptively adjusting the qualified device data according to the correlation between the applied parameters and the test results, the test scheme is optimized, which not only improves the accuracy of the test, but also shortens the test cycle and improves the overall work efficiency. In addition, by uploading the optimized test scheme to the test management platform, the test can be managed and executed automatically, which further improves the convenience and controllability of the test, ensures the consistency and traceability of the data during the test, and improves the production and testing quality of MOSFET devices. Overall, this method can effectively improve the performance evaluation efficiency, test accuracy and production qualification rate of MOSFET devices, reduce test cycles and costs, and optimize product quality.

[0054] Optionally, step S35 is specifically:

[0055] Step S351: performing device similarity calculation on the performance data of the devices that have passed the initial screening and the performance data of the devices that have failed the initial screening, thereby obtaining device similarity data;

[0056] Step S352: clustering similar devices on the performance data of the preliminarily screened qualified devices and the performance data of the preliminarily screened unqualified devices according to the device similarity data, thereby obtaining similar device clustering data;

[0057] Step S353: performing similar device performance error statistics on similar device clustering data, thereby obtaining low performance error similar device data;

[0058] Step S354: extracting low performance error similar device test application parameters from the historical test application parameter set according to the low performance error similar device data, thereby obtaining a low performance error similar device test application parameter set;

[0059] Step S355: Based on the application parameter-test result correlation data, the test application parameter set of the low performance error similar device test application parameter set is adaptively adjusted to obtain the MOSFET device test plan, and uploaded to the MOSFET device test management platform to execute the test plan implementation task.

[0060] The present invention can accurately identify devices with similar performance by calculating the similarity of performance data of qualified and unqualified devices in the initial screening, thereby providing a more representative data basis for subsequent cluster analysis. This process effectively reduces the differences between samples, making subsequent analysis more accurate and reliable. By clustering the devices by similarity, not only can the devices with similar performance be classified, but also the screening of the test objects can be optimized, making the test process more centralized and targeted, thereby saving test resources and improving work efficiency. By statistics on the performance errors of clustering data of similar devices, it is possible to timely find groups of devices with large performance deviations, so as to conduct deeper analysis and processing on these devices, and avoid affecting the overall test results due to the errors of a single device. Extracting the test application parameters of similar devices with low performance errors helps to apply specific test parameters to devices with performance problems. This process ensures the high pertinence and accuracy of the test scheme, and can optimize the characteristics of low-performance devices to avoid unnecessary waste of resources. Based on the correlation between the application parameters and the test results, the test parameters can be adaptively adjusted so that each group of similar devices with low performance errors can be tested under the most suitable conditions, further improving the accuracy and reliability of the test, and also optimizing the design of the test scheme. The combined effect of these steps not only improves the accuracy of MOSFET device performance testing, but also greatly improves the efficiency of the testing process, avoids excessive invalid tests, reduces manual intervention, and ultimately makes the test results more in line with actual needs, thereby promoting the improvement of overall device quality.

[0061] Optionally, step S5 specifically includes:

[0062] Performing a qualified device intersection operation on the performance data of the initially screened qualified devices and the performance data of the re-screened qualified devices, thereby obtaining the first screened qualified device data;

[0063] Perform device intersection operation on the performance data of the devices that failed the initial screening and the performance data of the devices that passed the rescreening, so as to obtain the screening error device data;

[0064] Calculating the performance difference of the screening error device based on the performance data of the primary screening unqualified device and the performance data of the re-screening qualified device, thereby obtaining the screening error device performance difference data;

[0065] Screening the screening error device data and removing high-performance difference devices according to the screening error device performance difference data, thereby obtaining second screening qualified device data;

[0066] The qualified device data of the first screening and the qualified device data of the second screening are summarized as qualified devices, so as to obtain a MOSFET device test report.

[0067] The present invention can obtain the first screening qualified device data by performing intersection operation on the primary screening qualified device data and the re-screening qualified device data, ensuring that only the devices that pass both screening stages will be considered qualified, thereby reducing the unqualified devices introduced due to single screening errors. Secondly, by performing intersection operation on the primary screening unqualified devices and the re-screening qualified devices, the screening error device data is obtained, which helps to identify and correct the erroneous data that may appear in the primary screening process, and provides a more accurate basis for further analysis. On this basis, by calculating the performance difference of the screening error device, the performance difference that may exist between the primary screening and the re-screening can be revealed, and the accuracy and consistency of the screening results can be further improved. The screening error device is screened out by high-performance difference devices, which can effectively exclude devices with large performance differences, thereby avoiding the interference of low-performance devices on subsequent tests and result analysis, and improving the quality of test data. Finally, by summarizing the first screening qualified device data and the second screening qualified device data, the generated MOSFET device test report can accurately and comprehensively reflect the performance of the device, and provide solid data support for subsequent decision-making and application. The entire process can ensure refined management of the screening process, thereby improving the accuracy, reliability and effectiveness of device testing, helping to improve product quality and reduce potential risks.

[0068] Optionally, the present specification also provides a rapid testing system for a MOSFET device, which is used to perform the rapid testing method for a MOSFET device as described above, and the rapid testing system for a MOSFET device includes:

[0069] The device IV characteristic integration module is used to obtain multi-source parallel data of MOSFET devices, and perform data preprocessing according to the multi-source parallel data of MOSFET devices, so as to obtain multi-source parallel data to be analyzed; and perform device IV characteristic integration on the multi-source parallel data to be analyzed, so as to obtain real-time device IV characteristic data;

[0070] The device performance prediction module is used to perform a voltage-current response analysis based on the real-time device IV characteristic data, thereby obtaining a real-time device voltage-current response curve; and to perform a MOSFET device performance prediction on the real-time device voltage-current response curve, thereby obtaining MOSFET device performance prediction data;

[0071] A test application parameter adjustment module is used to screen qualified devices according to the MOSFET device performance prediction data, so as to obtain the performance data of the preliminarily screened qualified devices; based on the preliminarily screened qualified device performance data, the test application parameters are adaptively adjusted to obtain the MOSFET device test plan, and uploaded to the MOSFET device test management platform to execute the test plan implementation task;

[0072] The device re-screening module is used to obtain the real-time test data of MOSFET devices, and to evaluate the device performance based on the real-time test data of MOSFET devices, thereby obtaining the device performance evaluation data; to screen qualified devices according to the device performance evaluation data, thereby obtaining the performance data of re-screened qualified devices;

[0073] The device test summary module is used to summarize the device tests based on the performance data of the devices that passed the initial screening and the performance data of the devices that passed the re-screening, so as to obtain a MOSFET device test report.

[0074] The rapid testing system of the MOSFET device of the present invention can implement any rapid testing method of the MOSFET device of the present invention, and is used to combine the operation and signal transmission medium between various modules to complete the rapid testing method of the MOSFET device. The internal modules of the system cooperate with each other, thereby greatly shortening the test cycle and reducing production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0076] Figure 1 A schematic flow chart of the steps of a rapid testing method for MOSFET devices of the present invention;

[0077] Figure 2 Detailed step flow diagram of step S1 in the present invention;

[0078] Figure 3 Detailed step flow diagram of step S2 in the present invention;

[0079] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0080] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0081] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0082] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0083] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for quickly testing a MOSFET device, the method comprising the following steps:

[0084] Step S1: acquiring multi-source parallel data of MOSFET devices, and performing data preprocessing according to the multi-source parallel data of MOSFET devices, thereby obtaining multi-source parallel data to be analyzed; integrating device IV characteristics of the multi-source parallel data to be analyzed, thereby obtaining real-time device IV characteristic data;

[0085] In this embodiment, different test equipment (such as source meter, digital oscilloscope, integrated test system, etc.) are used to test multiple MOSFET devices at the same time. These test equipment are connected to multiple MOSFET devices and record test data of multiple sources (such as temperature, bias voltage, gate voltage, etc.) under different working environments. The data collected by each device includes information such as voltage, current, temperature, etc., which comes from different test sources (such as different batches, different manufacturers, etc.) to form parallel data. The data obtained from different devices are preprocessed to ensure that the data format is unified and noise or invalid data is removed. Common data cleaning methods include removing missing data, filling blank values, data standardization, etc. By calculating the current-voltage (IV) characteristics of each MOSFET device, the IV characteristic curve of each MOSFET device is obtained based on multi-source parallel data. These curves will help subsequent performance evaluation and performance prediction. For example, data from different sources (different temperatures, gate voltages) are integrated into a unified IV characteristic diagram.

[0086] Step S2: performing a voltage-current response analysis based on the real-time device IV characteristic data to obtain a real-time device voltage-current response curve; performing a MOSFET device performance prediction on the real-time device voltage-current response curve to obtain MOSFET device performance prediction data;

[0087] In this embodiment, the obtained IV characteristic data is used to perform a voltage-current response analysis using numerical methods (such as least squares method, interpolation method, etc.). By adjusting the voltage range, calculating the current response and drawing the current-voltage curve, the response of the device under different voltages is obtained. During the analysis, the characteristics such as threshold voltage, on-current, and off-current are focused on, and the working state of the device is comprehensively evaluated in combination with variables such as temperature and gate voltage. Based on the voltage-current response curve, a machine learning algorithm (such as support vector machine, decision tree, etc.) is used to predict the performance of MOSFET devices. By training historical test data, the model can predict the performance of the device under different working conditions, such as maximum leakage current, switching speed, threshold voltage offset, etc.

[0088] Step S3: Screen qualified devices according to the MOSFET device performance prediction data, so as to obtain the performance data of the preliminarily screened qualified devices; adaptively adjust the test application parameters based on the preliminarily screened qualified device performance data, so as to obtain the MOSFET device test plan, and upload it to the MOSFET device test management platform to execute the test plan implementation task;

[0089] In this embodiment, according to the performance prediction results (such as current gain, leakage current, etc.), the MOSFET devices are screened using a preset performance threshold. For devices whose performance meets the requirements, they are marked as qualified devices, and their performance data (such as IV curve, power loss, etc.) are saved for subsequent analysis. For qualified devices that pass the preliminary screening, according to their performance data, an adaptive algorithm (such as fuzzy control, adaptive control, etc.) is used to adjust the application parameters of the test scheme. For example, the gate voltage application range is adjusted based on the threshold voltage of the device, or the test voltage is adjusted according to the maximum leakage current of the device to ensure that the test scheme can reflect the device performance to the greatest extent. The adjusted test scheme is uploaded to the MOSFET device test management platform. The platform will automatically schedule the corresponding test equipment according to the device status and execute the test steps defined in the scheme.

[0090] Step S4: acquiring real-time test data of MOSFET devices, and performing device performance evaluation based on the real-time test data of MOSFET devices, thereby obtaining device performance evaluation data; performing qualified device screening according to the device performance evaluation data, thereby obtaining re-screened qualified device performance data;

[0091] In this embodiment, during the retest process, the current, voltage, power loss and other data of the device are recorded in real time by connecting to the test system of the MOSFET device. During the test process, an automated system can be used to collect data, for example, a graphical user interface is used to display real-time waveforms and record data. According to the real-time test data, an evaluation algorithm (such as fuzzy logic evaluation or statistical analysis) is used to evaluate the performance of the MOSFET device. The evaluation content includes switching characteristics (such as switching time, rise time, fall time), threshold voltage stability, power efficiency, etc., to form performance evaluation data. According to the real-time test results, the performance threshold reflected by the initial screening qualified data is compared, and the devices that do not meet the requirements are eliminated. The re-screened qualified devices are screened out according to more stringent performance standards, and these data are further used for subsequent production and testing.

[0092] Step S5: Summarize the device tests based on the performance data of the devices that have passed the initial screening and the performance data of the devices that have passed the re-screening, so as to obtain a MOSFET device test report.

[0093] In this embodiment, the performance data of the qualified devices screened and rescreened are summarized to form a comprehensive test report. The report content includes the IV characteristics of each device, the performance prediction results, each key indicator in the test process (such as temperature, current, voltage, etc.), and the final performance evaluation results. The summarized data is analyzed and processed by the test report generation tool and displayed in the form of a visual chart or table. The report can also include some analysis conclusions, such as which performance parameters exceed the industry standards, which devices have potential quality problems, etc. The final generated MOSFET device test report will be uploaded to the product life cycle management platform for reference by the design team, quality control department, production team, etc., as a basis for further optimizing design and production.

[0094] Optionally, step S1 specifically includes:

[0095] Step S11: acquiring multi-source parallel data of MOSFET devices, and performing data preprocessing on the multi-source parallel data of MOSFET devices, thereby obtaining multi-source parallel data to be analyzed;

[0096] In this embodiment, performance data of MOSFET devices are collected from multiple sources. These data come from different test equipment, different experimental environments, or the same model of MOSFET devices from different manufacturing batches. For example, IV characteristic data of MOSFET can be collected in parallel by multiple current-voltage (IV) measuring instruments. Then, the raw data is processed using data preprocessing technology, including removing noise, filling missing data, standardizing data, etc. The key to data preprocessing is to ensure the accuracy and consistency of the data, such as using median filtering to remove high-frequency noise, or filling missing IV measurement points by interpolation to ensure the integrity and analyzability of the data.

[0097] Step S12: performing linear data division on the multi-source parallel data to be analyzed, thereby obtaining linear multi-source parallel data and nonlinear multi-source parallel data;

[0098] In this embodiment, the pre-processed data is divided into linear data and non-linear data. The IV characteristics of MOSFET usually show linear and non-linear characteristics in different working areas. For example, in the low voltage region (such as the subthreshold region), the IV characteristics of MOSFET show an exponential relationship, while in the saturation region, the IV characteristics show linearity. In order to complete the division of data, linearity and non-linearity can be distinguished by setting a threshold voltage range or fitting a specific voltage interval. For example, it can be set that when the voltage is less than a certain value, the fitting data is linear region data, and when the voltage is greater than the value, the fitting data is non-linear region data. In this way, the obtained linear multi-source parallel data and non-linear multi-source parallel data will be used for subsequent processing respectively.

[0099] Step S13: performing IV characteristic polynomial fitting on the linear multi-source parallel data to obtain linear IV characteristic data; performing IV characteristic exponential fitting on the nonlinear multi-source parallel data to obtain nonlinear IV characteristic data;

[0100] In this embodiment, IV characteristic polynomial fitting is performed on linear multi-source parallel data. In specific implementation, the relationship between current and voltage can be polynomially fitted using the least squares method, and quadratic or cubic polynomials are usually selected for fitting. For example, for IV data in the linear region, it can be fitted into a form similar to I=aV+b, where a and b are fitting coefficients. For nonlinear multi-source parallel data, exponential fitting is used, that is, an exponential function such as I=I_0 \exp(\frac{V}{nV_T})=I_0 \exp \left(\frac{V}{n V_{T}}\right) is used to fit the data, where I_0 is the initial value of the current, V_T is the thermal voltage, and n is the exponential fitting parameter. Through these fittings, linear and nonlinear IV characteristic data can be obtained respectively, which provide a basis for subsequent multi-source data merging.

[0101] Step S14: merging multi-source data based on the linear IV characteristic data and the nonlinear IV characteristic data to obtain device IV characteristic data;

[0102] In this embodiment, linear and nonlinear IV characteristic data obtained from different sources are merged to generate unified device IV characteristic data. The key to data merging is to ensure the coordination and consistency between different source data. For example, when merging, it is necessary to deal with problems such as unit differences and inconsistent scales in different data sources. A commonly used merging method is the weighted average method: different weights are assigned according to the credibility or accuracy of each data source, and then weighted average is performed. For example, some data sources may have higher measurement accuracy, so they can be assigned greater weights. Another method is to use interpolation technology to smoothly transition linear and nonlinear data in adjacent intervals. Through these merging strategies, the device IV characteristic data finally obtained can more comprehensively describe the current-voltage relationship of MOSFET devices under different operating conditions.

[0103] Step S15: performing real-time data integration on the multi-source parallel data to be analyzed according to the device IV characteristic data, thereby obtaining real-time device IV characteristic data.

[0104] In this embodiment, real-time data integration is performed on the multi-source parallel data to be analyzed based on the aforementioned device IV characteristic data. The goal of real-time data integration is to combine the newly collected data with the device IV characteristic data that has been analyzed to obtain real-time IV characteristic data. This usually involves online data processing and fusion techniques, such as Kalman filtering or particle filters, which are used to dynamically adjust and update data models. For example, when performing real-time monitoring, new IV data will be collected in real time and compared and corrected with previous data. By merging these new data, updated device IV characteristics can be obtained. This process not only ensures the timeliness of the data, but also provides continuous performance evaluation during equipment operation, helping to optimize the device's operating strategy.

[0105] Optionally, step S14 is specifically:

[0106] Step S141: dividing the linear IV characteristic data and the nonlinear IV characteristic data into acquisition time windows respectively, thereby obtaining windowed linear IV characteristic data and windowed nonlinear IV characteristic data;

[0107] In this embodiment, data acquisition is performed by an oscilloscope or a digital multimeter or other instrument. In order to effectively process these data, the collected IV data is divided into time windows, and the window size can usually be determined according to the sampling frequency of the measuring device. Assuming the sampling frequency is 1kHz, you can choose to collect 500 data points as a window each time. For linear IV characteristic data, the data segment can be identified and segmented by determining the linear relationship between current and voltage; for nonlinear IV characteristic data, the data window can be divided by finding obvious nonlinear intervals and using algorithms such as curve fitting. These steps will help us perform spectrum conversion more accurately in subsequent analysis.

[0108] Step S142: performing window spectrum conversion on the windowed linear IV characteristic data and the windowed nonlinear IV characteristic data respectively, so as to obtain a linear IV characteristic spectrum and a nonlinear IV characteristic spectrum;

[0109] In this embodiment, for the IV data in each time window, the fast Fourier transform (FFT) method can be used for spectrum conversion. Assuming that the window size is 500 sampling points, the data in each time window is first zero-filled to obtain a higher frequency resolution. Then, the time domain data is converted to the frequency domain using the FFT algorithm. For linear IV data, the converted spectrum shows that the main frequency components are concentrated in the lower frequency range; while for nonlinear IV data, more high-frequency components will appear in the spectrum, indicating the nonlinear characteristics of the system. The data after spectrum conversion can provide frequency domain information about the operating characteristics of the device, which is convenient for further analysis.

[0110] Step S143: Calculating characteristic spectrum similarity according to the linear IV characteristic spectrum and the nonlinear IV characteristic spectrum, thereby obtaining characteristic spectrum similarity data;

[0111] In this embodiment, a spectrum similarity calculation method is used to evaluate the similarity between linear and nonlinear IV characteristics. The specific implementation can adopt methods such as Pearson correlation coefficient, cosine similarity or dynamic time warping (DTW). For example, assuming that the linear IV characteristic spectrum is A and the nonlinear IV characteristic spectrum is B, the Pearson correlation coefficient between A and B can be calculated. If the correlation coefficient is close to 1, it means that the spectrum similarity between the two is high, otherwise it is low. In addition, a threshold can be set to divide the characteristic spectrum into two categories: similar and dissimilar according to the similarity result, providing a basis for the next step of characteristic segmentation merging.

[0112] Step S144: identifying characteristic segment merging points according to the characteristic spectrum similarity data, thereby obtaining characteristic segment merging point data;

[0113] In this embodiment, based on the characteristic spectrum similarity data, characteristic segmentation merging points can be identified, which represent the transition area from one characteristic segment to another. A specific implementation method is to set a similarity threshold (for example, 0.8) and identify the area with a similarity lower than the threshold as a merging point of the characteristic segment. For example, if the spectrum similarity value between two spectrum windows is lower than the set threshold, it can be determined that there is a significant change in the IV characteristics corresponding to the two windows, and it needs to be processed as a segmentation merging point. The identification of these merging points helps to perform more detailed segmentation of device characteristics in the future and improve the accuracy of the analysis results.

[0114] Step S145: performing characteristic segmentation merging on the linear IV characteristic data and the nonlinear IV characteristic data based on the characteristic segmentation merging point data, thereby obtaining device IV characteristic data.

[0115] In this embodiment, after identifying the characteristic segment merging points, these merging points are used as references to merge the linear IV characteristic data and the nonlinear IV characteristic data in segments. Assume that in the IV characteristic curve of a device, some areas show obvious linear behavior, while other areas show nonlinear behavior. After identifying the merging points, interpolation methods can be used to smoothly connect adjacent linear or nonlinear areas. For example, linear interpolation is used between linear areas, and polynomial interpolation is used for nonlinear areas. This processing method can effectively eliminate the breakpoints in the transition area, making the device IV characteristic data more continuous and smooth, which is convenient for subsequent analysis or modeling applications.

[0116] Optionally, step S2 specifically includes:

[0117] Step S21: constructing a voltage-current characteristic model based on real-time device IV characteristic data, thereby obtaining a voltage-current characteristic model;

[0118] In this embodiment, the real-time device IV characteristic data points are used as input, and a voltage-current characteristic model is constructed in combination with a regression analysis algorithm (such as least squares method, support vector machine regression, etc.). The purpose of this model is to accurately describe the current response of the MOSFET under given voltage conditions. In order to ensure the accuracy of the model, data needs to be collected under different environmental conditions (such as temperature and load changes), and polynomials or high-order nonlinear functions are used for fitting to ensure wide applicability and high accuracy.

[0119] Step S22: performing MOSFET device test mode recognition on the MOSFET device multi-source parallel data, thereby obtaining MOSFET device test mode data;

[0120] In this embodiment, data from multiple sources (for example, data from different test instruments, environmental conditions or operating frequencies) are unified and integrated. Data mining techniques (such as cluster analysis, K nearest neighbor algorithm, etc.) are used to perform pattern recognition on these data to identify different operating modes. These modes include normal operating mode, overload mode, saturation mode and linear mode, etc. The identification of the test mode is completed by comparing the changes in the IV characteristics of the device under different operating conditions. For example, under high voltage conditions, when the MOSFET enters the saturation region, the device current tends to be stable and can be identified as a saturation mode. Through pattern recognition, it is possible to accurately determine which operating state the current device is in, thereby providing a basis for subsequent simulation and prediction.

[0121] Step S23: performing a pole voltage mode simulation on the voltage-current characteristic model according to the MOSFET device test mode data, thereby obtaining pole voltage mode test simulation data;

[0122] In this embodiment, according to the obtained MOSFET test mode data, combined with the voltage-current characteristic model, simulation is performed in the extreme voltage mode. Specifically, the voltage range of the test (for example, the extreme value of the driving voltage) is first selected, and then the test mode is combined with the voltage-current characteristic model through simulation software (such as SPICE) to simulate the behavior of MOSFET in the extreme voltage mode. The goal of this step is to obtain simulation data for verifying the performance of the device under extreme conditions. During the test process, different load conditions and different voltage changes need to be selected to accurately reflect the response of the device under various emergency conditions.

[0123] Step S24: performing real-time device voltage-current response backtracking based on the pole voltage mode test simulation data, thereby obtaining a real-time device voltage-current response curve;

[0124] In this embodiment, based on the test simulation data in the extreme voltage mode, the voltage-current response curve of the real-time device is obtained by retrospective analysis technology (such as minimum error method, recursive neural network, etc.). First, the current response under different voltages is extracted from the simulation data, and then compared with the actual measured voltage-current characteristics to find the difference between the model and the real data. Through retrospective analysis, the voltage-current model is adjusted according to these differences so that it can more accurately simulate the real-time behavior of the device. In this process, multiple rounds of optimization are required, and real-time feedback data is relied on to continuously improve the prediction accuracy of the model.

[0125] Step S25: performing MOSFET device performance prediction on the real-time device voltage-current response curve, thereby obtaining MOSFET device performance prediction data.

[0126] In this embodiment, based on the voltage-current response curve obtained in step 1, a machine learning algorithm (such as random forest, neural network, support vector machine, etc.) is used to predict the performance of the MOSFET device. By using the voltage-current response curve as an input feature, the model predicts the key performance indicators of the device, such as switching speed, saturation current, leakage current, threshold voltage, etc. When training the model, a large amount of historical test data and the actual performance of the device are used as annotations to ensure that the model has a high prediction accuracy. For specific applications, the predicted data can be further compared with the preset performance requirements to evaluate whether the device meets the design requirements and optimize the subsequent manufacturing and testing processes.

[0127] Optionally, step S24 is specifically:

[0128] Performing voltage-current response curve conversion according to the pole voltage mode test simulation data to obtain a voltage-current simulation response curve, and performing frequency domain conversion on the voltage-current simulation response curve to obtain a pole voltage test response spectrum;

[0129] In this embodiment, the extreme voltage mode test simulation data is smoothed by a numerical algorithm (such as an interpolation method) to form a voltage-current response curve. Through this curve, the current response of the device at different voltage values ​​can be further derived to reveal its electrical performance. Assuming that the voltage in the extreme voltage mode test simulation data gradually increases from 0V to a certain value (such as 30V), the current response at each voltage point is extracted from the simulation data (for example, the current ranges from the microampere range to the milliampere range). These voltage-current data will draw a voltage-current response curve according to the voltage value (abscissa) and the corresponding current (ordinate). Assuming that the input voltage is 0V, 5V, 10V, and 20V, the measured current is 0μA, 10μA, 50μA, and 100μA, respectively, the voltage-current response curve finally drawn can reflect the behavior characteristics of the device at different voltages. The above voltage-current response curve is subjected to Fourier transform (FFT) to obtain the response spectrum in the frequency domain. Fourier transform converts time domain data into frequency domain, revealing the frequency components in voltage-current changes. Assuming that the sampling frequency of the voltage-current response data is 1 kHz, after FFT conversion, a spectrum is obtained, ranging from 0 Hz to 500 Hz. The analysis of this spectrum helps to identify characteristics such as high-frequency signals and low-frequency noise, and can effectively reveal the working status of the equipment and its performance anomalies.

[0130] Classify the frequency jitter characteristics according to the extreme voltage test response spectrum to obtain a smooth jitter spectrum and a sharp jitter spectrum;

[0131] In this embodiment, a signal processing algorithm is used to analyze the jitter characteristics in the spectrum. Steady and sharp frequency fluctuations are identified by calculating the amplitude changes in the spectrum. Assuming that the fluctuation of the signal is extracted through a frequency range with a bandwidth of 1kHz, if the amplitude changes of certain frequency bands in the spectrum are stable and the amplitude is small, it is marked as a "steady jitter spectrum"; if certain frequency bands have obvious amplitude fluctuations or periodic changes, they are marked as "sharp jitter spectrum". For example, in the spectrum, if the amplitude of a certain frequency band (such as 100Hz) rises and falls sharply, it may indicate sharp jitter.

[0132] Based on the stable jitter spectrum, the platform area curve of the voltage-current simulation response curve is identified to obtain the platform area response curve;

[0133] In this embodiment, the curve of the platform area of ​​the voltage-current response curve is identified based on the stable jitter spectrum. The platform area response curve is manifested as the stable part of the voltage-current curve, where the current is relatively stable and does not change much within a certain voltage range. Assuming that the voltage is between 0V and 20V, the current response changes little, forming a flat area, and this part of the data can be marked as the platform area response curve. For example, if the current is basically maintained at about 50μA in the range of 10V to 15V, it can be considered that this is a platform area response, indicating that the device is stable.

[0134] Based on the sharp jitter spectrum, the breakdown region curve is identified on the voltage-current simulation response curve, so as to obtain the breakdown region response curve;

[0135] In this embodiment, the sharp jitter spectrum represents the breakdown area of ​​the device, which is manifested as the sharp change part in the voltage-current response curve. When identifying the breakdown area curve, the response of the device in the case of rapid failure or sudden load change can be extracted by setting a threshold for rapid current change (for example, the current changes by more than a certain percentage in a short period of time). Assuming that when the voltage value is 3V, the current jumps from 1A to 3A in a very short period of time, it means that the device is overloaded or collapsed, which needs to be identified and analyzed later.

[0136] Performing current peak identification on the voltage-current simulation response curve to obtain current peak data;

[0137] In this embodiment, the current extreme point is identified by detecting the current peak in the voltage-current response curve. A peak detection algorithm can be used to first find the local maximum value in the current response curve. For example, when the voltage is between 10V and 30V, assuming that the current reaches a maximum value of 200μA at 20V in the current response curve, this point is the current peak. In some cases, it may be necessary to set a peak threshold to exclude irrelevant noise peaks.

[0138] The pole voltage and current characteristics are integrated according to the platform area response curve, the collapse area response curve and the current peak data, so as to obtain the pole voltage and current characteristic data;

[0139] In this embodiment, by combining the platform area response curve, the collapse area response curve and the current peak data, a voltage-current characteristic model of the device can be constructed. It is assumed that the platform area response curve represents the current-voltage relationship of the device under normal working conditions, the collapse area response curve represents the current-voltage change of the device under extreme load, and the current peak indicates the maximum carrying current of the device. Through the weighted average method or other multivariate analysis methods, the data of these three are integrated into a comprehensive characteristic data set to characterize the overall electrical performance of the device.

[0140] Through the voltage-current characteristic model, the real-time device voltage-current response is traced back to the real-time device IV characteristic data according to the pole voltage and current characteristic data, so as to obtain the real-time device voltage-current response curve.

[0141] In this embodiment, the IV characteristics of the real-time device are traced back through the obtained voltage-current characteristic model combined with the extreme voltage and current characteristic data. First, find the difference between the voltage-current characteristic model and the extreme voltage and current characteristic data, and adjust the voltage-current model according to these differences. The voltage-current characteristic data collected in real time is input into the voltage-current characteristic model combined with the extreme voltage and current characteristic data, and the actual current response curve of the device at the current voltage is obtained by back-calculation. Assuming that the voltage of a real-time device is 15V, by comparing the real-time current value at this voltage with the established voltage-current characteristic model, the current response of the device at this voltage can be predicted. For example, by backtracking, if the real-time current value is 80μA, this value can be used to confirm in the platform area or collapse area model to determine whether the device is in normal working state or close to a fault state.

[0142] Optionally, step S25 is specifically:

[0143] Performing breakdown region curve identification and platform region curve identification on the real-time device voltage-current response curve, thereby obtaining a real-time breakdown region curve and a real-time platform region curve;

[0144] In this embodiment, the real-time device voltage-current response curve obtained by analysis is used with a data processing algorithm (for example, slope analysis based on numerical differentiation or a detection method based on a threshold value) to identify the portion of the voltage-current curve where the current decreases rapidly as the collapse zone, while the current tends to be stable when the voltage is high, showing a stable platform area, that is, the platform area. To avoid noise interference, a current change rate threshold needs to be set. When the current change rate exceeds the set value, it is determined to be a collapse zone. For example, when the current drop rate exceeds 50A / V, it can be considered as a collapse zone, otherwise it is a platform zone.

[0145] Calculating the breakdown region curve ratio of the real-time device voltage-current response curve and the real-time breakdown region curve to obtain breakdown region curve ratio data;

[0146] In this embodiment, after obtaining the voltage-current response curve, the areas of the collapse region and the platform region are first separated from the complete voltage-current curve by image segmentation technology (such as a method based on data clustering). Then, the proportion of the voltage-current area of ​​the collapse region to the total response curve area is calculated. The specific implementation can obtain the collapse region curve proportion by calculating the ratio of the integral of the current corresponding to the collapse region voltage to the integral of the entire response curve area. For example, if the collapse region accounts for 30% in the voltage range, the collapse region proportion data is 30%.

[0147] Calculate the platform area curve ratio of the real-time device voltage-current response curve and the real-time platform area curve, so as to obtain platform area curve ratio data;

[0148] In this embodiment, similar to the calculation of the proportion of the collapse zone, the proportion of the platform area can be calculated by identifying the voltage-current interval of the platform area and calculating the ratio of its area to the area of ​​the total curve. The platform area is usually manifested as a stable area of ​​the current-voltage relationship, which can be confirmed by the current change rate being less than the set threshold. During the calculation, the voltage integral corresponding to the platform area current is compared with the integral of the total current response curve to obtain the proportion data of the platform area. For example, in the voltage-current response curve of the MOSFET device, the area of ​​the platform area is 0.6, and the area of ​​the total curve is 1, then the platform area accounts for 60%.

[0149] The device electrical performance is predicted based on the breakdown area curve proportion data and the platform area curve proportion data, thereby obtaining the MOSFET device performance prediction data.

[0150] In this embodiment, a performance prediction model can be established based on the percentage data of the collapse area and the platform area. This model can be trained by a machine learning algorithm (for example, a support vector machine (SVM) or a neural network), and the electrical performance indicators of the MOSFET device can be predicted by using the known percentage of the collapse area and the platform area of ​​the current response curve in the historical data. For example, through training data, indicators such as the breakdown voltage, on-resistance, and power loss of the MOSFET can be predicted. Assuming that the percentage of the collapse area is 30% and the percentage of the platform area is 70%, the performance of the MOSFET in long-term work can be predicted by a known training data model, such as the expected failure voltage or conduction loss of the device. Taking a 500V, 5A MOSFET device as an example, through real-time collection and analysis of its voltage-current response curve, a machine learning model is used to identify the collapse area and the platform area in the response curve. It is calculated that the area of ​​the collapse area accounts for 15%, the area of ​​the platform area accounts for 80%, and the rest is other areas. Based on these percentage data, the maximum operating power of the MOSFET device is 250W through regression analysis, and the working stability is high, which is suitable for high-frequency switching application scenarios.

[0151] Optionally, step S3 specifically includes:

[0152] Step S31: Calculating the electrical performance index of the MOSFET device according to the MOSFET device performance prediction data, thereby obtaining the MOSFET device prediction electrical performance index data;

[0153] In this embodiment, the performance index of the device is calculated using an electrical model according to the voltage-current response curve of the MOSFET device and the proportion data of the identified breakdown area and platform area. In specific implementation, it is assumed that the voltage response curve of the MOSFET presents certain nonlinear characteristics during the test process, the breakdown area indicates the failure area caused by the device being subjected to excessive current, and the platform area indicates that the device is in a stable working state. By calculating electrical performance indicators such as "maximum leakage current", "switching time", and "on-resistance" based on these curve areas, the predicted electrical performance data of the MOSFET device can be obtained. Taking "maximum leakage current" as an example, on the voltage-current response curve of the device, when the gate voltage gradually increases, the leakage current will show different growth trends. By calculating the relationship between current and voltage in real time, and combining the proportion of the breakdown area and the platform area, the maximum leakage current of the device under different gate voltages can be predicted. The maximum leakage current can be calculated on the voltage response curve according to the current change between the threshold value V_{th} of the gate voltage and the maximum operating voltage V_{max} by the formula I_{max}=\int_{V_{th}}^{V_{max}}I(V)dV. Similarly, the calculation of "switching time" and "on-resistance" also relies on the analysis of the voltage-current response curve. By monitoring the rate of change of the current response when the voltage changes rapidly (such as during switching), the response time of the MOSFET from "off" to "on" or reverse switching is determined. The switching time calculation can be quantified using the Rise Time and Fall Time, which can be obtained by numerically solving the integral value of the current change over time. For example, for the rise time, it can be calculated by the following steps: T_{rise}=t_{final}-t_{initial},\quad I(t_{initial})=0,\quad I(t_{final}) \text{reaches the maximum value}. The on-resistance is usually calculated by measuring the ratio of the voltage to the current when the MOSFET is in the on state. When the MOSFET is in the saturation region, the on-resistance can be expressed by the formula R_{on}=\frac{V_{DS}}{I_D}, where V_{DS} is the source-drain voltage and I_D is the drain current. For example, through model calculation, under a given operating voltage, if the MOSFET is in the platform area for a large proportion of the time, the device has good stability. Conversely, if the proportion of the breakdown area is too high, it indicates that the MOSFET has poor tolerance and is prone to failure.

[0154] Step S32: Screening the MOSFET device predicted electrical performance index data for qualified devices based on the preset MOSFET device qualified electrical performance index, thereby obtaining the performance data of the preliminarily screened qualified devices and the performance data of the preliminarily screened unqualified devices;

[0155] In this embodiment, the predicted electrical performance indicators of the MOSFET device are compared with the preset qualified standards (such as maximum leakage current, minimum switching time, etc.). In specific operation, according to the design requirements of the product, the qualified electrical performance standards are set as the following parameters: Maximum leakage current: for example, not more than 5 ; Minimum switching time: for example, not less than 50ns; Maximum on-resistance: for example, not more than 2 If the predicted electrical performance index meets the preset qualification standard, the device enters the qualified screening; if not, it is screened as unqualified. Assume that in a certain prediction, the maximum leakage current of MOSFET is 6 , and the default standard is 5 , the device is judged as unqualified and requires subsequent screening or processing.

[0156] Step S33: acquiring historical test data of the MOSFET device, and dividing the historical test data of the MOSFET device into test characteristics, thereby obtaining a historical test application parameter set and historical test result data;

[0157] In this embodiment, historical test data of MOSFET devices are extracted from the test database. Specifically, the historical test data includes MOSFET response characteristics under different voltage and current conditions, such as leakage current, switching time, stability, etc. It is necessary to perform feature division on these historical data. For example, a set of test application parameters (such as voltage and current) in the test data is extracted, and test results (such as leakage current and switching time) are extracted.

[0158] Step S34: Calculate the Spearman correlation coefficient for the historical test application parameter set and the historical test result data, so as to obtain application parameter-test result correlation data;

[0159] In this embodiment, the Spearman's rank correlation coefficient is used to evaluate the correlation between each applied parameter (such as voltage, current) and the test results (such as leakage current, switching time). Assuming that there is a significant correlation between certain applied parameters (such as voltage) and result data (such as switching time), this will help predict the performance of the device during the actual test process. For example, through Spearman's correlation calculation, it is found that the correlation coefficient between the test voltage and the MOSFET switching time is 0.85, indicating that the switching time may be prolonged when the voltage increases. These correlation data provide a basis for the subsequent optimization of the test scheme.

[0160] Step S35: Adaptively adjust the test application parameters of the initially screened qualified device performance data according to the application parameter-test result correlation data, thereby obtaining a MOSFET device test plan, and upload it to the MOSFET device test management platform to execute the test plan implementation task.

[0161] In this embodiment, the test scheme of the pre-screened qualified devices is adaptively adjusted according to the correlation data between the applied parameters and the test results. For example, if it is found that the switching time of some devices changes significantly when the voltage increases, the test scheme can be optimized to make the subsequent test conditions more accurate, so as to improve the test efficiency and reduce errors. Finally, the optimized test scheme is uploaded to the MOSFET device test management platform for execution. Assume that in an actual test scenario, it is found that the leakage current of the MOSFET is closely related to the applied current through Spearman correlation analysis. According to this data, the test scheme can be adjusted to focus on the change of leakage current when the current is high, so as to obtain a more accurate performance evaluation.

[0162] Optionally, step S35 is specifically:

[0163] Step S351: performing device similarity calculation on the performance data of the devices that have passed the initial screening and the performance data of the devices that have failed the initial screening, thereby obtaining device similarity data;

[0164] In this embodiment, the performance data of the MOSFET devices that are qualified and unqualified in the initial screening include but are not limited to parameters such as leakage current, switching speed, and temperature coefficient. Then, a similarity calculation method (such as Euclidean distance, cosine similarity, etc.) is used to compare these data to calculate the similarity between each device. The specific method can be to standardize the performance parameters of each device and obtain a similarity matrix through distance calculation to reflect the degree of performance similarity between the devices.

[0165] Step S352: clustering similar devices on the performance data of the preliminarily screened qualified devices and the performance data of the preliminarily screened unqualified devices according to the device similarity data, thereby obtaining similar device clustering data;

[0166] In this embodiment, based on the device similarity data obtained in 1, a clustering algorithm (such as K-means, hierarchical clustering, etc.) is used to classify the devices. During clustering, the devices are divided into different groups, each of which represents a set of devices with high similarity. For example, all devices with similarity greater than a certain threshold are clustered into one category, and vice versa. Devices with similarity can be effectively identified through this clustering process.

[0167] Step S353: performing similar device performance error statistics on similar device clustering data, thereby obtaining low performance error similar device data;

[0168] In this embodiment, the performance error analysis is performed on the cluster data of similar devices, and the performance error of the devices in each cluster during the test is calculated. For example, the error range between similar devices under different environments (such as temperature changes, voltage changes) is counted to determine which devices have smaller performance deviations, thereby identifying similar devices with smaller performance errors. At this time, the maximum tolerance value of the error can be set as the low performance judgment standard.

[0169] Step S354: extracting low performance error similar device test application parameters from the historical test application parameter set according to the low performance error similar device data, thereby obtaining a low performance error similar device test application parameter set;

[0170] In this embodiment, based on the determined low performance error similar devices, the test parameter set used in the historical test is selected, such as the test current, voltage, temperature and other environmental conditions, and the correlation between these parameters and the low performance error is analyzed to extract the parameter set that can optimize the performance error. For example, it is found that the performance fluctuation of a certain group of devices under a specific voltage is large, so the test voltage corresponding to this group of devices is extracted as the optimized application parameter set.

[0171] Step S355: Based on the application parameter-test result correlation data, the test application parameter set of the low performance error similar device test application parameter set is adaptively adjusted to obtain the MOSFET device test plan, and uploaded to the MOSFET device test management platform to execute the test plan implementation task.

[0172] In this embodiment, after obtaining a parameter set for testing similar devices with low performance errors, an adaptive adjustment algorithm (such as a genetic algorithm, a particle swarm optimization algorithm, etc.) is used to optimize the test parameters by analyzing the correlation between the parameters and the test results. For example, by optimizing the test range of voltage and temperature, the error fluctuation is reduced, thereby obtaining the optimal MOSFET device test scheme. Finally, the optimized test scheme is uploaded to the MOSFET device test management platform, and the actual test task is performed according to the scheme to ensure the accuracy and efficiency of the test results.

[0173] Optionally, step S5 specifically includes:

[0174] Performing a qualified device intersection operation on the performance data of the initially screened qualified devices and the performance data of the re-screened qualified devices, thereby obtaining the first screened qualified device data;

[0175] In this embodiment, the performance data of the devices that passed the initial screening and the performance data of the devices that passed the rescreening are respectively derived from two rounds of testing. By comparing the various electrical performance parameters of each device, such as leakage current (Idss), on-state voltage (Vth), maximum power, etc., an intersection operation is performed. Specifically, first, all devices that passed the initial screening are screened out, and their key performance parameters are recorded, and then the same parameters are recorded for the devices that passed the rescreening stage test. Through the intersection operation, the devices that meet the qualified standards in both the initial screening and the rescreening are finally screened out, that is, those devices that perform well in both rounds of testing and meet the design requirements. For example, assuming that the Vth of the devices that passed the initial screening is above 2V, and the Vth of the devices that passed the rescreening is required to be above 2.2V, then after the intersection operation, the devices with Vth above 2.2V will be retained as the first screened qualified devices.

[0176] Perform device intersection operation on the performance data of the devices that failed the initial screening and the performance data of the devices that passed the rescreening, so as to obtain the screening error device data;

[0177] In this embodiment, the unqualified devices in the initial screening are due to certain performance not meeting the specifications, such as excessive leakage current, threshold voltage deviation, etc. The qualified devices in the re-screening are devices that meet all requirements after more rigorous testing. At this time, by comparing the performance parameters of the unqualified devices in the initial screening with the qualified devices in the re-screening, the intersection operation can be used to obtain the screening error devices. These devices failed in the initial screening stage, but still performed well in the re-screening stage. Taking leakage current as an example, assuming that the leakage current of the device is greater than the specified threshold during the initial screening, and the leakage current of the re-screened device is smaller, the intersection operation will obtain an error device whose leakage current meets the requirements. Screening error device data can help further analyze the causes and make optimization measures.

[0178] Calculating the performance difference of the screening error device based on the performance data of the primary screening unqualified device and the performance data of the re-screening qualified device, thereby obtaining the screening error device performance difference data;

[0179] In this embodiment, the performance difference between the initial screening unqualified device and the re-screened qualified device can be quantified by calculating the performance difference of the screening error device. First, the performance data of the initial screening unqualified device and the performance data of the re-screened qualified device are obtained, and then the difference is calculated item by item, such as the difference between the leakage current, on-state voltage, power dissipation and other parameters. For the leakage current difference, assuming that the leakage current of the initial screening unqualified device is 3mA, and the leakage current of the re-screened qualified device is 1mA, the difference is 2mA. These performance difference data can help designers deeply analyze the source of errors, optimize the screening process, and improve the test method.

[0180] Screening the screening error device data and removing high-performance difference devices according to the screening error device performance difference data, thereby obtaining second screening qualified device data;

[0181] In this embodiment, by analyzing the performance difference data of the screened error devices, those error devices with large performance gaps can be identified, and these devices may have potential problems in quality control. Therefore, it is necessary to screen out these high-performance difference devices to improve the quality standard of the overall device. The specific method is to set a performance difference threshold, such as devices with a leakage current difference greater than 2mA are considered unqualified and screened out. In this way, those devices with too large differences can be accurately filtered out, and finally the second screened qualified device data that meets the requirements can be obtained.

[0182] The qualified device data of the first screening and the qualified device data of the second screening are summarized as qualified devices, so as to obtain a MOSFET device test report.

[0183] In this embodiment, the first screened qualified device data and the second screened qualified device data are summarized, and various performance indicators are integrated to compile a MOSFET device test report. The test report includes a summary of the test results of each batch of devices, a test parameter range, a qualified rate, a performance trend, and other contents. For example, the report will list the average value, maximum value, minimum value, and standard deviation of various key parameters such as threshold voltage (Vth), leakage current (Idss), power dissipation (Pd), and maximum operating frequency to ensure that the performance of each device meets the standard. At the same time, the report will also show the source of errors found during the test.

[0184] Optionally, the present specification also provides a rapid testing system for a MOSFET device, which is used to perform the rapid testing method for a MOSFET device as described above, and the rapid testing system for a MOSFET device includes:

[0185] The device IV characteristic integration module is used to obtain multi-source parallel data of MOSFET devices, and perform data preprocessing according to the multi-source parallel data of MOSFET devices, so as to obtain multi-source parallel data to be analyzed; and perform device IV characteristic integration on the multi-source parallel data to be analyzed, so as to obtain real-time device IV characteristic data;

[0186] The device performance prediction module is used to perform a voltage-current response analysis based on the real-time device IV characteristic data, thereby obtaining a real-time device voltage-current response curve; and to perform a MOSFET device performance prediction on the real-time device voltage-current response curve, thereby obtaining MOSFET device performance prediction data;

[0187] A test application parameter adjustment module is used to screen qualified devices according to the MOSFET device performance prediction data, so as to obtain the performance data of the preliminarily screened qualified devices; based on the preliminarily screened qualified device performance data, the test application parameters are adaptively adjusted to obtain the MOSFET device test plan, and uploaded to the MOSFET device test management platform to execute the test plan implementation task;

[0188] The device re-screening module is used to obtain the real-time test data of MOSFET devices, and to evaluate the device performance based on the real-time test data of MOSFET devices, thereby obtaining the device performance evaluation data; to screen qualified devices according to the device performance evaluation data, thereby obtaining the performance data of re-screened qualified devices;

[0189] The device test summary module is used to summarize the device tests based on the performance data of the devices that passed the initial screening and the performance data of the devices that passed the re-screening, so as to obtain a MOSFET device test report.

[0190] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0191] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be 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 the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A fast testing method for MOSFET devices, characterized in that: The following steps are involved: Step S1: acquiring multi-source parallel data of MOSFET devices, and performing data preprocessing according to the multi-source parallel data of MOSFET devices, thereby obtaining multi-source parallel data to be analyzed; Integrate device IV characteristics of multi-source parallel data to be analyzed to obtain real-time device IV characteristics data; Step S2: performing a voltage-current response analysis based on the real-time device IV characteristic data, thereby obtaining a real-time device voltage-current response curve; Predict MOSFET device performance based on real-time device voltage-current response curve, thereby obtaining MOSFET device performance prediction data; Step S3: Screen qualified devices according to the MOSFET device performance prediction data, so as to obtain the performance data of the preliminarily screened qualified devices; adaptively adjust the test application parameters based on the preliminarily screened qualified device performance data, so as to obtain the MOSFET device test plan, and upload it to the MOSFET device test management platform to execute the test plan implementation task; The test application parameters are a set of test application parameters including voltage and current from historical test data of MOSFET devices extracted from a preset test database; Step S4: acquiring real-time test data of the MOSFET device, and performing device performance evaluation based on the real-time test data of the MOSFET device, thereby obtaining device performance evaluation data; Qualified devices are screened according to device performance evaluation data, thereby obtaining performance data of re-screened qualified devices; Step S5: Summarize the device tests based on the performance data of the devices that have passed the initial screening and the performance data of the devices that have passed the re-screening, so as to obtain a MOSFET device test report.

2. The fast testing method of MOSFET device according to claim 1, characterized in that: Step S1 is specifically as follows: Step S11: acquiring multi-source parallel data of MOSFET devices, and performing data preprocessing on the multi-source parallel data of MOSFET devices, thereby obtaining multi-source parallel data to be analyzed; Step S12: performing linear data division on the multi-source parallel data to be analyzed, thereby obtaining linear multi-source parallel data and nonlinear multi-source parallel data; Step S13: performing IV characteristic polynomial fitting on the linear multi-source parallel data to obtain linear IV characteristic data; performing IV characteristic exponential fitting on the nonlinear multi-source parallel data to obtain nonlinear IV characteristic data; Step S14: merging multi-source data based on the linear IV characteristic data and the nonlinear IV characteristic data to obtain device IV characteristic data; Step S15: performing real-time data integration on the multi-source parallel data to be analyzed according to the device IV characteristic data, thereby obtaining real-time device IV characteristic data.

3. The rapid testing method for MOSFET device according to claim 2, characterized in that: Step S14 is specifically as follows: Step S141: dividing the linear IV characteristic data and the nonlinear IV characteristic data into acquisition time windows respectively, thereby obtaining windowed linear IV characteristic data and windowed nonlinear IV characteristic data; Step S142: performing window spectrum conversion on the windowed linear IV characteristic data and the windowed nonlinear IV characteristic data respectively, so as to obtain a linear IV characteristic spectrum and a nonlinear IV characteristic spectrum; Step S143: Calculating characteristic spectrum similarity according to the linear IV characteristic spectrum and the nonlinear IV characteristic spectrum, thereby obtaining characteristic spectrum similarity data; Step S144: identifying characteristic segment merging points according to the characteristic spectrum similarity data, thereby obtaining characteristic segment merging point data; Step S145: performing characteristic segmentation merging on the linear IV characteristic data and the nonlinear IV characteristic data based on the characteristic segmentation merging point data, thereby obtaining device IV characteristic data.

4. The rapid testing method for MOSFET device according to claim 1, characterized in that: Step S2 is specifically as follows: Step S21: constructing a voltage-current characteristic model based on real-time device IV characteristic data, thereby obtaining a voltage-current characteristic model; Step S22: performing MOSFET device test mode recognition on the MOSFET device multi-source parallel data, thereby obtaining MOSFET device test mode data; Step S23: performing a pole voltage mode simulation on the voltage-current characteristic model according to the MOSFET device test mode data, thereby obtaining pole voltage mode test simulation data; Step S24: performing real-time device voltage-current response backtracking based on the pole voltage mode test simulation data, thereby obtaining a real-time device voltage-current response curve; Step S25: performing MOSFET device performance prediction on the real-time device voltage-current response curve, thereby obtaining MOSFET device performance prediction data.

5. The rapid testing method of MOSFET device according to claim 1, characterized in that: Step S24 is specifically as follows: Performing voltage-current response curve conversion according to the pole voltage mode test simulation data to obtain a voltage-current simulation response curve, and performing frequency domain conversion on the voltage-current simulation response curve to obtain a pole voltage test response spectrum; Classify the frequency jitter characteristics according to the extreme voltage test response spectrum to obtain a smooth jitter spectrum and a sharp jitter spectrum; Based on the stable jitter spectrum, the platform area curve of the voltage-current simulation response curve is identified to obtain the platform area response curve; Based on the sharp jitter spectrum, the breakdown region curve is identified on the voltage-current simulation response curve, so as to obtain the breakdown region response curve; Performing current peak identification on the voltage-current simulation response curve to obtain current peak data; The pole voltage and current characteristics are integrated according to the platform area response curve, the collapse area response curve and the current peak data, so as to obtain the pole voltage and current characteristic data; Through the voltage-current characteristic model, the real-time device voltage-current response is traced back to the real-time device IV characteristic data according to the pole voltage and current characteristic data, so as to obtain the real-time device voltage-current response curve.

6. The rapid testing method for MOSFET device according to claim 5, characterized in that: Step S25 is specifically as follows: Performing breakdown region curve identification and platform region curve identification on the real-time device voltage-current response curve, thereby obtaining a real-time breakdown region curve and a real-time platform region curve; Calculating the breakdown region curve ratio of the real-time device voltage-current response curve and the real-time breakdown region curve to obtain breakdown region curve ratio data; Calculate the platform area curve proportion of the real-time device voltage-current response curve and the real-time platform area curve, so as to obtain platform area curve proportion data; The device electrical performance is predicted based on the breakdown area curve proportion data and the platform area curve proportion data, thereby obtaining the MOSFET device performance prediction data.

7. The rapid testing method of MOSFET device according to claim 1, characterized in that: Step S3 is specifically as follows: Step S31: Calculating the electrical performance index of the MOSFET device according to the MOSFET device performance prediction data, thereby obtaining the MOSFET device prediction electrical performance index data; Step S32: Screening the MOSFET device predicted electrical performance index data for qualified devices based on the preset MOSFET device qualified electrical performance index, thereby obtaining the performance data of the preliminarily screened qualified devices and the performance data of the preliminarily screened unqualified devices; Step S33: acquiring historical test data of the MOSFET device, and dividing the historical test data of the MOSFET device into test characteristics, thereby obtaining a historical test application parameter set and historical test result data; Step S34: Calculate the Spearman correlation coefficient for the historical test application parameter set and the historical test result data, so as to obtain application parameter-test result correlation data; Step S35: Adaptively adjust the test application parameters of the initially screened qualified device performance data according to the application parameter-test result correlation data, thereby obtaining a MOSFET device test plan, and upload it to the MOSFET device test management platform to execute the test plan implementation task.

8. The rapid testing method for MOSFET device according to claim 7, characterized in that: Step S35 is specifically as follows: Step S351: performing device similarity calculation on the performance data of the devices that have passed the initial screening and the performance data of the devices that have failed the initial screening, thereby obtaining device similarity data; Step S352: clustering similar devices on the performance data of the devices that have passed the initial screening and the performance data of the devices that have failed the initial screening according to the device similarity data, thereby obtaining similar device clustering data; Step S353: performing similar device performance error statistics on similar device clustering data, thereby obtaining low performance error similar device data; Step S354: extracting low performance error similar device test application parameters from the historical test application parameter set according to the low performance error similar device data, thereby obtaining a low performance error similar device test application parameter set; Step S355: Based on the application parameter-test result correlation data, the test application parameter set of the low performance error similar device test application parameter set is adaptively adjusted to obtain the MOSFET device test plan, and uploaded to the MOSFET device test management platform to execute the test plan implementation task.

9. The fast testing method of MOSFET device according to claim 1, characterized in that: Step S5 is specifically as follows: Performing a qualified device intersection operation on the performance data of the initially screened qualified devices and the performance data of the re-screened qualified devices, thereby obtaining the first screened qualified device data; Perform device intersection operation on the performance data of the devices that failed the initial screening and the performance data of the devices that passed the rescreening, so as to obtain the screening error device data; Calculating the performance difference of the screening error device based on the performance data of the primary screening unqualified device and the performance data of the re-screening qualified device, thereby obtaining the screening error device performance difference data; Screening the screening error device data and removing high-performance difference devices according to the screening error device performance difference data, thereby obtaining second screening qualified device data; The qualified device data of the first screening and the qualified device data of the second screening are summarized as qualified devices, so as to obtain a MOSFET device test report.

10. A fast testing system for MOSFET devices, characterized in that: A method for rapidly testing a MOSFET device according to claim 1, wherein the rapid testing system comprises: The device IV characteristic integration module is used to obtain multi-source parallel data of MOSFET devices, and perform data preprocessing according to the multi-source parallel data of MOSFET devices, so as to obtain multi-source parallel data to be analyzed; and perform device IV characteristic integration on the multi-source parallel data to be analyzed, so as to obtain real-time device IV characteristic data; The device performance prediction module is used to perform a voltage-current response analysis based on the real-time device IV characteristic data, thereby obtaining a real-time device voltage-current response curve; and to perform a MOSFET device performance prediction on the real-time device voltage-current response curve, thereby obtaining MOSFET device performance prediction data; A test application parameter adjustment module is used to screen qualified devices according to the MOSFET device performance prediction data, so as to obtain the performance data of the preliminarily screened qualified devices; based on the preliminarily screened qualified device performance data, the test application parameters are adaptively adjusted to obtain the MOSFET device test plan, and uploaded to the MOSFET device test management platform to execute the test plan implementation task; The device re-screening module is used to obtain the real-time test data of MOSFET devices, and to evaluate the device performance based on the real-time test data of MOSFET devices, thereby obtaining the device performance evaluation data; to screen qualified devices according to the device performance evaluation data, thereby obtaining the performance data of re-screened qualified devices; The device test summary module is used to summarize the device tests based on the performance data of the devices that passed the initial screening and the performance data of the devices that passed the re-screening, so as to obtain a MOSFET device test report.

Images