Data-driven frequency converter switching device aging analysis method and system
By real-time monitoring and multi-dimensional aging correlation enhancement of inverter switching devices, combined with a multi-physics field coupling analytical model, the problem of lagging aging status assessment of inverter switching devices is solved, and real-time and accurate aging status assessment and maintenance management are realized.
Patent Information
- Application Number
- CN202511887275.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-15
- Publication Date
- 2026-03-17
AI Technical Summary
The aging status assessment of existing inverter switching devices relies on periodic inspections and experience-based judgments, resulting in a lag in assessments. This makes it impossible to reflect the actual aging status of the devices in real time and lacks a comprehensive, multi-dimensional assessment, which affects the operating efficiency and safety of the equipment.
By real-time monitoring of inverter switching devices, a multi-dimensional aging correlation enhancement vector is obtained. The coupling effect is compensated using a multi-physics field coupling analytical model. Combined with an adversarial sample generator and a deep neural network, a multi-dimensional aging state assessment is performed, generating an aging assessment map for maintenance and management.
It enables real-time and accurate assessment of the aging status of inverter switching devices, improving the accuracy and real-time performance of the assessment and ensuring the safe and reliable operation of the equipment.
Smart Images
Figure CN121679308A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of electrical measurement technology, specifically to a data-driven method and system for aging analysis of inverter switching devices. Background Technology
[0002] Current assessments of the aging status of inverter switching devices typically rely on periodic inspections and experience-based judgment. This introduces a lag in the assessment process, failing to reflect the actual aging condition of the devices in real time. Traditional assessment methods often provide only single-dimensional data analysis, lacking a comprehensive evaluation of the multi-dimensional aging status of devices. In practical applications, these methods cannot accurately determine the degree of device aging, leading to the failure to promptly detect potential faults and impacting equipment operating efficiency and safety. Summary of the Invention
[0003] This application provides a data-driven method and system for aging analysis of inverter switching devices, which addresses the technical problems of inaccurate aging status assessment and poor real-time performance of inverter switching devices in the prior art.
[0004] In view of the above problems, this application provides a data-driven method and system for aging analysis of inverter switching devices.
[0005] The first aspect of this application provides a data-driven method for aging analysis of inverter switching devices, the method comprising: Real-time monitoring of the inverter switching devices is performed to obtain a device monitoring dataset. Based on the device monitoring dataset, multi-dimensional aging correlation attention enhancement is applied to the inverter switching devices to obtain a multi-dimensional aging correlation enhancement vector. Based on the multi-dimensional aging correlation enhancement vector, a multi-dimensional aging state assessment is performed on the inverter switching devices to obtain a first aging assessment map. Based on the inverter state data, transient impact suppression correction is applied to the first aging assessment map to obtain a second aging assessment map. A multi-physics coupling analytical model is introduced to compensate for the coupling effect of the second aging assessment map to obtain a third aging assessment map. Maintenance and management of the inverter switching devices are performed based on the third aging assessment map.
[0006] A second aspect of this application provides a data-driven aging analysis system for inverter switching devices, the system comprising: The system includes: a real-time monitoring module for real-time monitoring of inverter switching devices to obtain a device monitoring dataset; an attention enhancement module for performing multi-dimensional aging-related attention enhancement on the inverter switching devices based on the device monitoring dataset to obtain a multi-dimensional aging-related enhancement vector; an aging state assessment module for performing multi-dimensional aging state assessment on the inverter switching devices based on the multi-dimensional aging-related enhancement vector to obtain an aging assessment first spectrum; a correction module for performing transient impact suppression correction on the aging assessment first spectrum based on inverter state data to obtain an aging assessment second spectrum; a coupling effect compensation module for introducing a multi-physics coupling analytical model to compensate for the coupling effect on the aging assessment second spectrum to obtain an aging assessment third spectrum; and a maintenance management module for performing maintenance management on the inverter switching devices based on the aging assessment third spectrum.
[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application performs real-time monitoring of inverter switching devices to obtain a device monitoring dataset; based on the device monitoring dataset, it performs multi-dimensional aging correlation attention enhancement on the inverter switching devices to obtain a multi-dimensional aging correlation enhancement vector; based on the multi-dimensional aging correlation enhancement vector, it performs multi-dimensional aging state assessment on the inverter switching devices to obtain an aging assessment first spectrum; based on the inverter state data, it performs transient impact suppression correction on the aging assessment first spectrum to obtain an aging assessment second spectrum; it introduces a multi-physics coupling analytical model to compensate for the coupling effect of the aging assessment second spectrum to obtain an aging assessment third spectrum; and it performs maintenance management on the inverter switching devices based on the aging assessment third spectrum. This invention solves the technical problems of inaccurate and poor real-time performance in the aging state assessment of inverter switching devices in the prior art, achieving improved assessment accuracy and real-time performance through real-time monitoring and multi-dimensional aging correlation enhancement. Attached Figure Description
[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0009] Figure 1 A schematic flowchart of a data-driven aging analysis method for inverter switching devices provided in this application embodiment; Figure 2 A schematic diagram of the structure of a data-driven inverter switching device aging analysis system provided in this application embodiment.
[0010] Figure labeling: Real-time monitoring module 11, Attention enhancement module 12, Aging status assessment module 13, Correction module 14, Coupling effect compensation module 15, Maintenance management module 16. Detailed Implementation
[0011] This application provides a data-driven method and system for aging analysis of inverter switching devices. It addresses the technical problems of inaccurate and poor real-time assessment of the aging status of inverter switching devices in the prior art by providing a data-driven method and system for aging analysis of inverter switching devices. Through real-time monitoring and multi-dimensional aging correlation enhancement, it achieves the technical effect of improving the accuracy and real-time performance of the assessment.
[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.
[0013] It should be noted that any variation of the terms "comprising" and "having" is intended to cover non-exclusive inclusion, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such processes, methods, products, or devices.
[0014] Example 1, as Figure 1 As shown, this application provides a data-driven method for aging analysis of inverter switching devices, the method comprising: Step S100: Perform real-time monitoring of the inverter switching devices to obtain device monitoring dataset.
[0015] In this embodiment, the inverter switching devices are monitored in real time using pre-deployed current sensors, voltage sensors, temperature sensors, and frequency sensors. The current sensors measure the current flowing through the switching devices, the voltage sensors monitor voltage changes in real time, the temperature sensors detect temperature fluctuations, and the vibration sensors capture vibration signals that may be caused by mechanical failures or aging. This real-time monitoring process yields a device monitoring dataset. This dataset includes multi-dimensional data on the inverter switching devices, such as current, voltage, and temperature, at various times.
[0016] Step S200: Perform multidimensional aging-related attention enhancement on the inverter switching device based on the device monitoring dataset to obtain a multidimensional aging-related enhancement vector.
[0017] In this embodiment, when performing multidimensional aging-related attention enhancement on inverter switching devices based on device monitoring datasets, the inverter switching devices are first analyzed in terms of structure, performance, and lifespan aging. For structural aging, features related to structural changes are extracted and enhanced to obtain a structural aging enhancement vector. For performance aging, features related to device performance degradation are analyzed and enhanced to obtain a performance aging enhancement vector. For lifespan aging, features related to device lifespan are analyzed and enhanced to obtain a lifespan aging enhancement vector. Finally, the structural aging enhancement vector, performance aging enhancement vector, and lifespan aging enhancement vector are merged to form a multidimensional aging correlation vector.
[0018] Furthermore, in the method provided in the application embodiments, the process of performing multi-dimensional aging-related attention enhancement on the inverter switching device based on the device monitoring dataset to obtain a multi-dimensional aging-related enhancement vector further includes: Based on the device monitoring dataset, structural aging correlation attention enhancement is performed on the inverter switching device to obtain a structural aging correlation enhancement vector; based on the device monitoring dataset, performance aging correlation attention enhancement is performed on the inverter switching device to obtain a performance aging correlation enhancement vector; based on the device monitoring dataset, lifetime aging correlation attention enhancement is performed on the inverter switching device to obtain a lifetime aging correlation enhancement vector; the structural aging correlation enhancement vector, the performance aging correlation enhancement vector, and the lifetime aging correlation enhancement vector are encapsulated into the multidimensional aging correlation enhancement vector.
[0019] In this embodiment, firstly, for structural aging correlation enhancement, feature cleaning is performed on the device monitoring dataset to filter out key features related to device structural changes, such as current and voltage fluctuations. The cleaned data forms a device monitoring feature set. Then, structural aging correlation evaluation is performed on each device monitoring feature in the feature set to identify features related to structural aging, forming a feature structural aging correlation set. Finally, attention enhancement processing is performed on the feature structural aging correlation set to enhance features closely related to structural aging, ultimately generating a structural aging correlation enhancement vector.
[0020] Similarly, for performance aging association enhancement, the performance-related features in the device monitoring dataset are first cleaned to obtain a performance feature set. Next, performance aging association evaluation is performed on these feature sets to identify features associated with performance degradation, generating a feature performance aging association set. Then, attention enhancement processing is used to highlight the features associated with performance aging, forming a performance aging association enhancement vector.
[0021] Similarly, for lifetime aging correlation enhancement, lifetime-related features in the monitoring dataset are cleaned to form a lifetime feature set. Then, lifetime aging correlation evaluation is performed on these features to identify those related to device lifetime aging, generating a feature lifetime aging correlation set. Next, attention enhancement processing is performed to obtain a lifetime aging correlation enhancement vector, reflecting the device's lifetime aging state.
[0022] Finally, the structural aging correlation enhancement vector, performance aging correlation enhancement vector, and lifetime aging correlation enhancement vector are encapsulated to obtain a multidimensional aging correlation enhancement vector.
[0023] Furthermore, in the method provided in the application embodiments, the structural aging correlation attention enhancement is performed on the inverter switching device based on the device monitoring dataset to obtain a structural aging correlation enhancement vector, and the method further includes: Feature cleaning is performed on the device monitoring dataset to obtain a device monitoring feature set; structural aging association evaluation is performed on each device monitoring feature in the device monitoring feature set to obtain a feature structural aging association set; attention enhancement processing is performed on the device monitoring feature set based on the feature structural aging association set to generate the structural aging association enhancement vector.
[0024] In this embodiment, when cleaning features based on the device monitoring dataset, a mean imputation method is used to handle missing data. In this process, for each monitoring feature, such as current, voltage, and temperature, missing records in the dataset are first identified. For each feature with missing values, the mean of all valid data for that feature is calculated. Then, the calculated mean is used to impute the missing portions. Through this process, the cleaned data constitutes the device monitoring feature set.
[0025] Next, a structural aging correlation evaluation is performed on each device monitoring feature within the device monitoring feature set. In this process, the Pearson correlation coefficient is used to quantify the linear correlation between each feature and structural aging. During the evaluation, if the Pearson correlation coefficient is greater than 0.7, a strong positive correlation between the feature and structural aging is considered; if the correlation coefficient is less than -0.7, a strong negative correlation is indicated. By calculating the correlation coefficient between each feature and structural aging, features with a strong linear relationship to structural aging are selected, and a feature-structural aging correlation set is generated.
[0026] Finally, based on the feature-structure aging correlation set, attention enhancement processing is applied to the device monitoring feature set. This step employs a self-attention mechanism, dynamically adjusting the influence of each feature according to its correlation with structural aging. Specifically, the self-attention mechanism calculates the relationship between each feature and other features, and enhances the contribution of related features based on their influence on structural aging. For example, if the current feature has a strong correlation with structural aging (e.g., a correlation coefficient of 0.85), the self-attention mechanism will increase the weight of that feature in the evaluation. Conversely, for features with weak correlation (e.g., the temperature feature with a correlation coefficient of 0.5), its influence on the evaluation results is reduced. After self-attention enhancement processing, a structural aging correlation enhancement vector is generated.
[0027] Step S300: Perform a multi-dimensional aging status assessment on the inverter switching device based on the multi-dimensional aging correlation enhancement vector to obtain the first aging assessment map.
[0028] In this embodiment, when assessing the multidimensional aging state of inverter switching devices based on multidimensional aging correlation enhancement vectors, a global aging assessment is first performed on the structural aging of the inverter switching devices to obtain a first record set for structural aging assessment. Then, an adversarial sample generator is used to inject perturbations into the first record set to obtain a second record set for structural aging assessment. Subsequently, adversarial training is performed based on the second record set to obtain a structural aging assessment model. This structural aging assessment model is then used to perform assessments based on the structural aging correlation enhancement vectors to obtain the structural aging assessment results.
[0029] Next, an adversarial example generator is used to assess the aging state of the performance aging correlation enhancement vector and the lifetime aging correlation enhancement vector, respectively, to obtain the performance aging assessment results and the lifetime aging assessment results. Finally, the structural aging assessment results and the performance aging assessment results are combined to generate the first aging assessment map.
[0030] Furthermore, in the method provided in the application embodiments, the multi-dimensional aging state assessment of the inverter switching device is performed based on the multi-dimensional aging correlation enhancement vector to obtain an aging assessment first spectrum, which further includes: A global aging assessment record set for structural aging is obtained by performing a global search on the inverter switching devices. A second record set for structural aging is obtained by injecting perturbations into the first record set using an adversarial example generator. An adversarial training model is obtained based on the second record set. A structural aging assessment result is obtained based on the structural aging correlation enhancement vector and the structural aging assessment model. An aging state assessment is performed on the performance aging correlation enhancement vector using the adversarial example generator to obtain a performance aging assessment result. An aging state assessment is performed on the lifetime aging correlation enhancement vector using the adversarial example generator to obtain a lifetime aging assessment result. Finally, the first aging assessment map is generated by combining the structural aging assessment result and the performance aging assessment result.
[0031] In this embodiment, a global aging assessment of the inverter switching devices is first performed. During this process, a global retrieval algorithm, such as the k-NN algorithm based on similarity metrics, is used to retrieve data on multiple inverter switching devices that are similar to the target inverter switching device in terms of structural aging from a pre-defined database. The historical data of these inverter switching devices constitute the first record set for structural aging assessment.
[0032] Next, an adversarial example generator is used to inject perturbations into the obtained first set of structural aging assessment records. In this process, the adversarial example generator adds small perturbations to each data point in the first set of structural aging assessment records to simulate the changes that the inverter switching devices may undergo under different operating conditions, thereby obtaining the second set of structural aging assessment records.
[0033] Subsequently, adversarial training was performed on the second record set for structural aging assessment using a deep neural network model. In this process, both disturbed and normal samples were input into the deep neural network model for training. The model learned patterns and features in the data by optimizing the loss function. This adversarial training enabled the deep neural network model to accurately assess the structural aging state of the frequency converter even when faced with disturbances and noise. Through this process, the final structural aging assessment model was obtained.
[0034] Subsequently, based on the structural aging assessment model and the structural aging correlation enhancement vector, a structural aging assessment of the inverter switching devices is performed. By inputting the structural aging correlation enhancement vector into the structural aging assessment model, the model performs forward propagation based on the previously trained features to generate the structural aging assessment result. This assessment result reflects the overall state of the inverter switching devices in terms of structural aging.
[0035] The following steps are similar to the performance aging and lifespan aging assessment process. First, a global search for performance aging assessment is performed. A similarity metric algorithm is used to retrieve performance aging data of the same model of inverter switching devices from the database, generating a first record set for performance aging assessment. Then, an adversarial example generator is used to inject perturbations into the first record set, generating a second record set for performance aging assessment. Similarly, adversarial training is performed on the generated second record set to obtain the performance aging assessment model. Finally, based on the performance aging association enhancement vector, the performance aging assessment result is obtained using the performance aging assessment model.
[0036] For life aging assessment, a global aging assessment search is first performed to collect life aging data of switching devices of the same model of frequency converter, generating a first record set for life aging assessment. Then, an adversarial example generator is used to inject perturbations into the first record set for life aging assessment, generating a second record set for life aging assessment. After adversarial training, a life aging assessment model is obtained. This model can assess the life of frequency converter switching devices based on life aging correlation enhancement vectors and generate life aging assessment results.
[0037] Finally, the structural aging assessment results, performance aging assessment results, and lifespan aging assessment results are integrated. Through data fusion methods, such as multidimensional fusion technology, an aging assessment first map is generated. This map comprehensively displays the assessment results of the inverter switching devices in different aging dimensions.
[0038] Step S400: Perform transient impact suppression correction on the first aging assessment spectrum based on the inverter status data to obtain the second aging assessment spectrum.
[0039] In this embodiment, when performing transient impact suppression correction on the first aging assessment spectrum based on inverter status data, inverter status data is first collected through a real-time monitoring system. This data includes key operating parameters such as current, voltage, temperature, and frequency. Next, using this data, transient impact detection is performed on the inverter's switching devices to identify drastic changes in current and voltage over a short period, yielding device impact detection results. Subsequently, based on the detected transient impacts, the effects on structural aging, performance aging, and lifespan aging are analyzed, resulting in transient impact characteristics for structural aging, performance aging, and lifespan aging. Finally, these impact characteristics are combined to adaptively correct the first aging assessment spectrum, removing interference from transient impacts on the aging status assessment and generating the second aging assessment spectrum.
[0040] Furthermore, in the method provided in the application embodiment, the transient impact suppression correction is performed on the first aging assessment spectrum based on the inverter status data to obtain the second aging assessment spectrum, which further includes: Transient impact detection is performed on the inverter switching devices based on the inverter status data to obtain device impact detection results; structural aging transient impact identification is performed based on the device impact detection results to obtain structural aging transient impact characteristics; performance aging transient impact identification is performed based on the device impact detection results to obtain performance aging transient impact characteristics; lifetime aging transient impact identification is performed based on the device impact detection results to obtain lifetime aging transient characteristics; adaptive correction is performed on the first aging assessment map based on the structural aging transient impact characteristics, the performance aging transient impact characteristics, and the lifetime aging transient characteristics to generate the second aging assessment map.
[0041] In this embodiment, transient impact detection is first performed on the monitoring signal using a threshold detection method based on the inverter's status data. Specifically, the real-time sequences of current, voltage, temperature, and mechanical vibration are compared with set thresholds. When the rate of change of voltage or current exceeds the electrical impact threshold, the temperature experiences a rapid rise or fall exceeding the thermal impact threshold within a short period, or the acceleration of mechanical vibration exceeds the mechanical impact threshold, these are respectively labeled as electrical impact, thermal impact, and mechanical impact. Through this step, the device impact detection results are obtained, which record the occurrence time, intensity, and corresponding parameters of different types of transient impacts.
[0042] Next, based on the device impact test results, a peak detection method is used to identify the transient effects of structural aging. In this process, peak values are extracted from the current waveform, temperature curve, and vibration curve before and after the impact. The peak value difference and duration are calculated to identify indicators related to material fatigue, solder joint damage, and mechanical loosening. For example, if the peak value difference before and after the current impact is greater than a preset reference value, it is determined to have a significant impact on the structure. Through this step, the transient impact characteristics of structural aging are obtained, which quantitatively characterize the impact of transient impacts on device structural aging.
[0043] Subsequently, based on the device impact test results, a linear regression method was used to identify the transient effects of performance aging. In this process, the transient impact intensity was used as the independent variable, and power output, current fluctuations, or efficiency degradation as the dependent variable. A linear regression equation was constructed, and the regression coefficients and coefficients of determination were calculated to evaluate the correlation between transient impacts and performance degradation. For example, when the regression coefficient is positive and significant, it indicates that the higher the impact intensity, the more severe the performance degradation. Through this step, the transient impact characteristics of performance aging were obtained, which are used to quantitatively reflect the degree to which performance aging is affected by transient impacts.
[0044] Subsequently, based on the device impact test results, the Weibull lifetime analysis method was used to identify the transient effects of lifetime aging. In this process, the number and intensity of impacts were used as input conditions to fit the historical lifetime distribution of the inverter switching devices. The shape and dimensional parameters of impacted and non-impacted samples were compared to calculate the lifetime reduction rate. For example, a significant decrease in dimensional parameters indicates a significant impact of impact on lifetime shortening. Through this step, transient characteristics of lifetime aging were obtained, revealing the contributions of electrical, thermal, and mechanical shocks to lifetime degradation.
[0045] Finally, based on the transient impact characteristics of structural aging, performance aging, and lifespan aging, a moving average method is used to adaptively correct the first aging assessment spectrum. Specifically, fluctuation points containing transient impacts are smoothed to remove short-term anomalies and retain long-term trends, thereby eliminating the bias caused by transient impacts on the spectrum. Through this step, a second aging assessment spectrum is generated, which, after removing the influence of transient impacts, more accurately reflects the long-term aging state of the inverter's switching devices.
[0046] Step S500: Introduce a multi-physics field coupling analytical model to compensate for the coupling effect of the second aging assessment spectrum, and obtain the third aging assessment spectrum.
[0047] In this embodiment, when using a multiphysics coupled analytical model to compensate for the coupling effect of the second aging assessment map, firstly, a multidimensional aging path tracing is performed based on the second aging assessment map, dividing the aging evolution process into structural aging path, performance aging path, and lifetime aging path. Then, a multiphysics coupled analytical model is introduced to superimpose factors such as electric field, thermal field, and force field onto each type of aging path. Compensation is applied to the structural aging path to obtain the first device aging path. Compensation is applied to the performance aging path to obtain the second device aging path. Compensation is applied to the lifetime aging path to obtain the third device aging path. Finally, the first, second, and third device aging paths are fused to generate the third aging assessment map.
[0048] Furthermore, in the method provided in the application embodiments, a multiphysics coupling analytical model is introduced to compensate for the coupling effect of the second aging assessment spectrum to obtain the third aging assessment spectrum, which further includes: Based on the second aging assessment map, a multi-dimensional aging path tracing is performed to determine the structural aging path, performance aging path, and lifetime aging path. The structural aging path is then compensated for coupling effects using the multiphysics coupling analytical model to obtain the first device aging path. The performance aging path is then compensated for coupling effects using the same model to obtain the second device aging path. The lifetime aging path is then compensated for coupling effects using the same model to obtain the third device aging path. Finally, the third aging assessment map is generated based on the first, second, and third device aging paths.
[0049] In this embodiment of the application, a multi-dimensional aging path tracing is first performed on the second aging assessment map. By employing a time series decomposition method, the changing trends of different dimensions in the second aging assessment map are separated, and curves related to structural degradation, performance degradation, and lifespan degradation are extracted respectively to form structural aging path, performance aging path, and lifespan aging path.
[0050] After obtaining the structural aging path, the coupling effect is compensated based on a multiphysics coupling analytical model. Specifically, multiphysics parameters are collected from the inverter switching devices to obtain a multi-field monitoring sequence including electric, thermal, and force fields. This multi-field monitoring sequence is then input into the multiphysics coupling analytical model, which outputs the analytical results of the multi-field coupling effect. Based on the analytical results, multi-node aging effect identification is performed to obtain a coupling effect aging effect map. The structural aging path is then corrected based on this map, ultimately forming the first aging path of the device.
[0051] Subsequently, the compensation process for the performance aging path and the lifetime aging path is consistent with the compensation method for the structural aging path. For the performance aging path, multi-field monitoring sequences are input and processed through a multi-physics coupled analytical model. Combined with influence identification and path correction, a second device aging path is obtained, which characterizes the correction results of performance degradation under multi-field effects. For the lifetime aging path, a third device aging path is obtained through multi-physics parameter acquisition, analytical result calculation, aging influence identification, and path correction, which describes the compensation situation of lifetime degradation under multi-field conditions.
[0052] Finally, the first, second, and third aging paths of the device are merged to generate a third aging assessment map. Supported by a multiphysics coupling analytical model, this third aging assessment map integrates the path compensation results of structural aging, performance aging, and lifespan aging, comprehensively depicting the true aging state of the inverter switching devices under multi-field coupling.
[0053] Furthermore, in the method provided in the application embodiments, the coupling effect compensation is performed on the structural aging path according to the multiphysics coupling analytical model to obtain the first aging path of the device, which further includes: Multi-physics field parameters are acquired for the inverter switching devices to obtain a multi-field monitoring sequence; the multi-field monitoring sequence is input into the multi-physics field coupling analytical model to obtain the multi-field coupling effect analysis result; based on the multi-field coupling effect analysis result, the aging path of the structure is identified by multi-node aging influence to obtain the coupling effect aging influence map; based on the coupling effect aging influence map, the aging path of the structure is corrected to obtain the first aging path of the device.
[0054] In this embodiment, when acquiring multi-physics field parameters of the inverter switching devices, voltage sensors, current sensors, thermocouple temperature sensors, and acceleration sensors are used to continuously monitor the voltage, current, temperature, and mechanical vibration of the inverter switching devices during operation, and these monitoring signals are recorded in the form of a time series. The electrical signal sequence, thermal signal sequence, and mechanical signal sequence formed by the sensor acquisition constitute a multi-field monitoring sequence.
[0055] Next, the multi-field monitoring sequences are input into a multi-physics coupling analytical model. This model is built using machine learning methods. Principal component analysis is used to reduce the dimensionality of the collected multi-field monitoring sequences, extracting key features characterizing the coupling effects of voltage, current, temperature, and acceleration. Then, a support vector machine is used to train the multi-dimensional features, obtaining the mapping relationships between the electric, thermal, and force fields. By inputting the multi-field monitoring sequences into the trained multi-physics coupling analytical model, the analytical results of the multi-field coupling effect are obtained. These results describe the mutual influence between the field variables under the combined action of multiple physical fields.
[0056] Subsequently, based on the analysis results of multi-field coupling effects, the aging path of the structure was identified through multi-node aging impact identification. In this process, peak detection methods and time interval division were used to divide the structural aging path into several nodes. Each node was then compared with the analysis results of multi-field coupling effects to extract the response characteristics of each node under the combined action of electrical stress, thermal stress, and mechanical stress. Through node-by-node comparison and calculation, a coupling effect aging impact map covering the entire structural aging path was formed.
[0057] Finally, based on the aging effect spectrum of the coupling effect, the structural aging path is corrected. The correction step uses the moving average method and curve fitting method to correct the node data affected by multiple fields such as electricity, heat, and force, and remove the deviation caused by a single field factor. The path after correction can accurately reflect the real aging trend under the action of multiple physics fields, and finally obtain the first aging path of the device.
[0058] Step S600: Perform maintenance and management on the inverter switching devices according to the third aging assessment map.
[0059] In this embodiment, when maintaining and managing the inverter switching devices based on the third aging assessment map, a threshold determination method is used to compare the structural aging characteristics, performance aging characteristics, and lifespan aging characteristics extracted from the third aging assessment map with preset standard values. When the structural aging path exceeds the set limit, structural component maintenance is performed; when the performance aging path shows efficiency or power deviations, performance parameters are adjusted or components are replaced; when the lifespan aging path is below the remaining lifespan threshold, preventative maintenance or complete machine replacement is arranged. In this way, structural maintenance instructions, performance maintenance instructions, and lifespan maintenance instructions are integrated into a maintenance management scheme, thereby achieving maintenance management based on the third aging assessment map and ensuring the reliable operation of the inverter switching devices.
[0060] Furthermore, the method provided in the application embodiments also includes: Based on the impact detection results of the device, a transient impact warning signal is generated.
[0061] In this embodiment of the application, based on the device impact detection results, when an electrical impact, thermal impact, or mechanical impact is detected, a transient impact warning signal is generated and alerted through an alarm device to achieve real-time safety warning for the inverter switching devices.
[0062] Furthermore, the method provided in the application embodiments also includes: Based on the third aging assessment spectrum, a device aging warning signal is generated.
[0063] In this embodiment, based on the third aging assessment map, when the evolution trend of the structural aging path, performance aging path, or lifetime aging path indicates that the device aging state has reached the warning condition, the aging characteristics of the corresponding node are extracted, and a device aging warning signal is generated. This warning signal serves as the output of the maintenance and control process and is alerted through an alarm device.
[0064] In summary, the embodiments of this application have at least the following technical effects: This application performs real-time monitoring of inverter switching devices to obtain a device monitoring dataset; based on the device monitoring dataset, it performs multi-dimensional aging correlation attention enhancement on the inverter switching devices to obtain a multi-dimensional aging correlation enhancement vector; based on the multi-dimensional aging correlation enhancement vector, it performs multi-dimensional aging state assessment on the inverter switching devices to obtain an aging assessment first spectrum; based on the inverter state data, it performs transient impact suppression correction on the aging assessment first spectrum to obtain an aging assessment second spectrum; it introduces a multi-physics coupling analytical model to compensate for the coupling effect of the aging assessment second spectrum to obtain an aging assessment third spectrum; and it performs maintenance management on the inverter switching devices based on the aging assessment third spectrum. This invention solves the technical problems of inaccurate and poor real-time performance in the aging state assessment of inverter switching devices in the prior art, achieving improved assessment accuracy and real-time performance through real-time monitoring and multi-dimensional aging correlation enhancement.
[0065] Example 2, based on the same inventive concept as the data-driven aging analysis method for inverter switching devices in the foregoing examples, such as... Figure 2 As shown, this application provides a data-driven aging analysis system for inverter switching devices. The system and method embodiments in this application are based on the same inventive concept. The system includes: The system comprises: a real-time monitoring module 11 for real-time monitoring of inverter switching devices to obtain a device monitoring dataset; an attention enhancement module 12 for performing multi-dimensional aging-related attention enhancement on the inverter switching devices based on the device monitoring dataset to obtain a multi-dimensional aging-related enhancement vector; an aging state assessment module 13 for performing multi-dimensional aging state assessment on the inverter switching devices based on the multi-dimensional aging-related enhancement vector to obtain an aging assessment first spectrum; a correction module 14 for performing transient impact suppression correction on the aging assessment first spectrum based on inverter state data to obtain an aging assessment second spectrum; a coupling effect impact compensation module 15 for introducing a multi-physics field coupling analytical model to compensate for the coupling effect impact on the aging assessment second spectrum to obtain an aging assessment third spectrum; and a maintenance management module 16 for performing maintenance management on the inverter switching devices based on the aging assessment third spectrum.
[0066] Furthermore, the system is also used to implement the following functions: Based on the device monitoring dataset, structural aging correlation attention enhancement is performed on the inverter switching device to obtain a structural aging correlation enhancement vector; based on the device monitoring dataset, performance aging correlation attention enhancement is performed on the inverter switching device to obtain a performance aging correlation enhancement vector; based on the device monitoring dataset, lifetime aging correlation attention enhancement is performed on the inverter switching device to obtain a lifetime aging correlation enhancement vector; the structural aging correlation enhancement vector, the performance aging correlation enhancement vector, and the lifetime aging correlation enhancement vector are encapsulated into the multidimensional aging correlation enhancement vector.
[0067] Furthermore, the system is also used to implement the following functions: Feature cleaning is performed on the device monitoring dataset to obtain a device monitoring feature set; structural aging association evaluation is performed on each device monitoring feature in the device monitoring feature set to obtain a feature structural aging association set; attention enhancement processing is performed on the device monitoring feature set based on the feature structural aging association set to generate the structural aging association enhancement vector.
[0068] Furthermore, the system is also used to implement the following functions: A global aging assessment record set for structural aging is obtained by performing a global search on the inverter switching devices. A second record set for structural aging is obtained by injecting perturbations into the first record set using an adversarial example generator. An adversarial training model is obtained based on the second record set. A structural aging assessment result is obtained based on the structural aging correlation enhancement vector and the structural aging assessment model. An aging state assessment is performed on the performance aging correlation enhancement vector using the adversarial example generator to obtain a performance aging assessment result. An aging state assessment is performed on the lifetime aging correlation enhancement vector using the adversarial example generator to obtain a lifetime aging assessment result. Finally, the first aging assessment map is generated by combining the structural aging assessment result and the performance aging assessment result.
[0069] Furthermore, the system is also used to implement the following functions: Transient impact detection is performed on the inverter switching devices based on the inverter status data to obtain device impact detection results; structural aging transient impact identification is performed based on the device impact detection results to obtain structural aging transient impact characteristics; performance aging transient impact identification is performed based on the device impact detection results to obtain performance aging transient impact characteristics; lifetime aging transient impact identification is performed based on the device impact detection results to obtain lifetime aging transient characteristics; adaptive correction is performed on the first aging assessment map based on the structural aging transient impact characteristics, the performance aging transient impact characteristics, and the lifetime aging transient characteristics to generate the second aging assessment map.
[0070] Furthermore, the system is also used to implement the following functions: Based on the second aging assessment map, a multi-dimensional aging path tracing is performed to determine the structural aging path, performance aging path, and lifetime aging path. The structural aging path is then compensated for coupling effects using the multiphysics coupling analytical model to obtain the first device aging path. The performance aging path is then compensated for coupling effects using the same model to obtain the second device aging path. The lifetime aging path is then compensated for coupling effects using the same model to obtain the third device aging path. Finally, the third aging assessment map is generated based on the first, second, and third device aging paths.
[0071] Furthermore, the system is also used to implement the following functions: Multi-physics field parameters are acquired for the inverter switching devices to obtain a multi-field monitoring sequence; the multi-field monitoring sequence is input into the multi-physics field coupling analytical model to obtain the multi-field coupling effect analysis result; based on the multi-field coupling effect analysis result, the aging path of the structure is identified by multi-node aging influence to obtain the coupling effect aging influence map; based on the coupling effect aging influence map, the aging path of the structure is corrected to obtain the first aging path of the device.
[0072] Furthermore, the system is also used to implement the following functions: Based on the impact detection results of the device, a transient impact warning signal is generated.
[0073] Furthermore, the system is also used to implement the following functions: Based on the third aging assessment spectrum, a device aging warning signal is generated.
[0074] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.
[0075] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.
[0076] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.
Claims
1. A data-driven inverter switching device aging analysis method, characterized by, The method comprises: Real-time monitoring of the frequency converter switching device to obtain a device monitoring data set; Multi-dimensional aging correlation attention enhancement of the frequency converter switching device according to the device monitoring data set to obtain a multi-dimensional aging correlation enhancement vector; Multi-dimensional aging state evaluation of the frequency converter switching device according to the multi-dimensional aging correlation enhancement vector to obtain an aging evaluation first graph; Transient impact suppression correction of the aging evaluation first graph according to the frequency converter state data to obtain an aging evaluation second graph; Coupling effect influence compensation of the aging evaluation second graph by introducing a multi-physical field coupling analytical model to obtain an aging evaluation third graph; Maintenance management of the frequency converter switching device according to the aging evaluation third graph.
2. The data-driven inverter switching device aging analysis method of claim 1, wherein, The multi-dimensional aging correlation attention enhancement of the frequency converter switching device according to the device monitoring data set to obtain a multi-dimensional aging correlation enhancement vector comprises: Structural aging correlation attention enhancement of the frequency converter switching device according to the device monitoring data set to obtain a structural aging correlation enhancement vector; Performance aging correlation attention enhancement of the frequency converter switching device according to the device monitoring data set to obtain a performance aging correlation enhancement vector; Service life aging correlation attention enhancement of the frequency converter switching device according to the device monitoring data set to obtain a service life aging correlation enhancement vector; The structural aging correlation enhancement vector, the performance aging correlation enhancement vector, and the service life aging correlation enhancement vector are packaged into the multi-dimensional aging correlation enhancement vector.
3. The data-driven inverter switching device aging analysis method of claim 2, wherein, The structural aging correlation attention enhancement of the frequency converter switching device according to the device monitoring data set to obtain a structural aging correlation enhancement vector comprises: Feature cleaning according to the device monitoring data set to obtain a device monitoring feature set; Structural aging correlation evaluation of each device monitoring feature in the device monitoring feature set to obtain a feature structural aging correlation set; Attention enhancement processing of the device monitoring feature set according to the feature structural aging correlation set to generate the structural aging correlation enhancement vector.
4. The method of claim 1, wherein the data-driven inverter switching device aging analysis is performed by a computer system. The multi-dimensional aging state evaluation of the frequency converter switching device according to the multi-dimensional aging correlation enhancement vector to obtain an aging evaluation first graph comprises: Structural aging evaluation global retrieval of the frequency converter switching device to obtain a structural aging evaluation first record set; Disturbance injection of the structural aging evaluation first record set according to an adversarial sample generator to obtain a structural aging evaluation second record set; Adversarial training according to the structural aging evaluation second record set to obtain a structural aging evaluation model; Structural aging evaluation results based on the structural aging correlation enhancement vector according to the structural aging evaluation model; Performance aging correlation enhancement vector aging state evaluation according to the adversarial sample generator to obtain performance aging evaluation results; Service life aging correlation enhancement vector aging state evaluation according to the adversarial sample generator to obtain service life aging evaluation results, and the structural aging evaluation results and the performance aging evaluation results are combined to generate the aging evaluation first graph.
5. The method of claim 1, wherein the data-driven inverter switching device aging analysis is performed by a computer system. According to the variable frequency converter state data, the first aging evaluation graph is corrected to resist transient impact, and a second aging evaluation graph is obtained, including: According to the variable frequency converter state data, the variable frequency converter switching device is detected for transient impact, and a device impact detection result is obtained; According to the device impact detection result, structural aging transient impact identification is performed, and a structural aging transient impact feature is obtained; According to the device impact detection result, performance aging transient impact identification is performed, and a performance aging transient impact feature is obtained; According to the device impact detection result, life aging transient impact identification is performed, and a life aging transient feature is obtained; According to the structural aging transient impact feature, the performance aging transient impact feature and the life aging transient feature, the first aging evaluation graph is adaptively corrected to generate the second aging evaluation graph.
6. The method of claim 1, wherein the data-driven inverter switching device aging analysis is performed by a computer system. A multi-physical field coupling analysis model is introduced to compensate the coupling effect influence of the second aging evaluation graph, and a third aging evaluation graph is obtained, including: According to the second aging evaluation graph, a multi-dimensional aging path is traced to determine a structural aging path, a performance aging path and a life aging path; According to the multi-physical field coupling analysis model, the structural aging path is compensated for coupling effect influence to obtain a first device aging path; According to the multi-physical field coupling analysis model, the performance aging path is compensated for coupling effect influence to obtain a second device aging path; According to the multi-physical field coupling analysis model, the life aging path is compensated for coupling effect influence to obtain a third device aging path; According to the first device aging path, the second device aging path and the third device aging path, the third aging evaluation graph is generated.
7. The method of claim 6, wherein the data-driven inverter switching device aging analysis is performed by a computer system. According to the multi-physical field coupling analysis model, the structural aging path is compensated for coupling effect influence to obtain a first device aging path, including: The variable frequency converter switching device is collected for multi-physical field parameters to obtain a multi-field monitoring sequence; The multi-field monitoring sequence is input into the multi-physical field coupling analysis model to obtain a multi-field coupling effect analysis result; According to the multi-field coupling effect analysis result, multi-node aging impact identification is performed on the structural aging path to obtain a coupling effect aging impact graph; According to the coupling effect aging impact graph, the structural aging path is corrected to obtain the first device aging path.
8. The method of claim 5, wherein the data-driven inverter switching device aging analysis is performed by a computer system. According to the device impact detection result, a transient impact early warning signal is generated.
9. The data-driven inverter switching device aging analysis method of claim 1, wherein, According to the third aging evaluation graph, a device aging early warning signal is generated.
10. A data-driven inverter switching device aging analysis system, characterized by, The system is used to perform the data-driven variable frequency converter switching device aging analysis method as claimed in any one of claims 1-9, and the system includes: A real-time monitoring module for real-time monitoring of the variable frequency converter switching device to obtain a device monitoring data set; An attention enhancement module for multi-dimensional aging correlation attention enhancement of the variable frequency converter switching device according to the device monitoring data set to obtain a multi-dimensional aging correlation enhancement vector; An aging state evaluation module for multi-dimensional aging state evaluation of the variable frequency converter switching device according to the multi-dimensional aging correlation enhancement vector to obtain a first aging evaluation graph; The correction module is configured to correct the first aging evaluation graph according to the frequency converter state data to obtain a second aging evaluation graph; The coupling effect influence compensation module is configured to introduce a multi-physical field coupling analysis model to compensate the second aging evaluation graph according to the coupling effect influence to obtain a third aging evaluation graph; The maintenance management module is configured to perform maintenance management on the frequency converter switching device according to the third aging evaluation graph.