A power capacitor multifunctional detection system and method

Abstract

This invention relates to the field of power testing technology and proposes a multifunctional testing system and method for power capacitors. The system includes an access state baseline calibration module, an electrothermal stress superposition excitation module, a charge / discharge characteristic trajectory tracking module, a dielectric polarization segmented slope identification module, a multidimensional fault feature weighted fusion module, and a closed-loop detection interactive feedback execution module. The system performs baseline calibration on the state monitoring signal of the capacitor under test to obtain the basic electrical test environment; based on this environment, it applies electrothermal stress superposition excitation to the capacitor to obtain the environmental response signal; it performs constant current source switching and voltage trajectory tracking on the response signal to obtain the charge / discharge characteristic curve; it performs segmented slope identification on the charge / discharge characteristic curve to obtain the internal dielectric polarization characteristics; it weightedly fuses these characteristics with the ripple component to generate a multidimensional fault diagnosis vector; and it achieves closed-loop detection of the capacitor through interactive feedback. This invention can improve the efficiency of multifunctional testing of power capacitors.

Description

Technical Field

[0001] This invention relates to the field of power testing technology, and in particular to a multifunctional testing system and method for power capacitors. Background Technology

[0002] Existing power capacitor testing technologies struggle to achieve accurate baseline calibration for the connected state, and incomplete signal interference filtering can lead to insufficient stability in the testing environment, failing to provide a reliable foundation for subsequent testing. Furthermore, the simplistic electrothermal stress excitation method cannot create a highly adaptable composite excitation, resulting in low accuracy in acquiring capacitor environmental response signals and an inability to accurately reflect the equipment's operating status.

[0003] Traditional detection methods lack sufficient accuracy in tracking charge and discharge characteristics, lack segmented analytical capabilities for identifying dielectric polarization features, and suffer from low efficiency in fault feature extraction and fusion. They also fail to generate accurate multi-dimensional diagnostic vectors and lack closed-loop interactive feedback mechanisms, resulting in reliability and efficiency that are insufficient for practical applications. Therefore, improving the accuracy and intelligence of power capacitor detection has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a multifunctional testing system and method for power capacitors to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a multifunctional detection system for power capacitors, characterized in that the system includes an access state baseline calibration module, an electrothermal stress superposition excitation module, a charge / discharge characteristic trajectory tracking module, a dielectric polarization segmented slope identification module, a multidimensional fault feature weighted fusion module, and a closed-loop detection interactive feedback execution module, wherein: The access status baseline calibration module acquires the access status monitoring signal of the capacitor under test and performs baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test. The electrothermal stress superposition excitation module, based on the basic electrical test environment, performs electrothermal stress superposition excitation on the capacitor under test to obtain the environmental response signal of the capacitor under test; The charge / discharge characteristic trajectory tracking module switches the constant current source of the environmental response signal to obtain the initial voltage value of the capacitor under test, and performs voltage trajectory tracking on the initial voltage value to obtain the charge / discharge characteristic curve of the capacitor under test. The dielectric polarization segmented slope identification module identifies the segmented slope of the charging saturation region and the discharging cutoff region in the charging and discharging characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test. The multidimensional fault feature weighted fusion module performs feature weighted fusion of the internal dielectric polarization features and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test. The closed-loop detection interactive feedback execution module provides interactive feedback to the multi-dimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test.

[0006] In a preferred embodiment, when the access status baseline calibration module acquires the access status monitoring signal of the capacitor under test and performs baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test, it is specifically used for: Acquire the connection status monitoring signal of the capacitor under test; The DC component of the access status monitoring signal is filtered out to obtain the DC-free signal of the capacitor under test; The frequency domain characteristic spectrum of the capacitor under test is obtained by performing spectral analysis on the DC component signal. Based on the energy distribution density in the frequency domain characteristic spectrum, the strong interference frequency band of the capacitor under test is identified to generate the adaptive notch filter parameters of the capacitor under test. Based on the adaptive notch filter parameters, the DC component removal signal is iteratively filtered to obtain the pure timing signal of the capacitor under test. The steady-state reference range of the capacitor under test is obtained by performing sliding window variance statistics on the pure time series signal. Based on the steady-state reference range, zero-point drift compensation is performed on the pure timing signal to obtain the basic electrical test environment of the capacitor under test.

[0007] In a preferred embodiment, when the electrothermal stress superposition excitation module performs electrothermal stress superposition excitation on the capacitor under test based on the basic electrical test environment to obtain the environmental response signal of the capacitor under test, it is specifically used for: The background impedance data in the basic electrical test environment is processed by time-frequency domain transformation to obtain the load characteristic vector of the capacitor under test; Based on a preset multi-dimensional excitation strategy library, the load characteristic vector is indexed and matched to obtain the excitation parameters of the capacitor under test. The dynamic excitation parameters are time-series segmented and recombined to obtain the composite excitation sequence of the capacitor under test; Based on the composite excitation sequence, the capacitor under test is subjected to segmented data-driven excitation to obtain the original response data stream of the capacitor under test; The original response data stream is subjected to sliding window filtering to obtain the environmental response signal of the capacitor under test.

[0008] In a preferred embodiment, the charge / discharge characteristic trajectory tracking module, when performing constant current source switching on the environmental response signal to obtain the initial voltage value of the capacitor under test, is specifically used for: The environmental response signal is time-series segmented to obtain a multidimensional data segment of the capacitor under test. By performing correlation feature matching on the multidimensional data fragments, the feature marker points of the capacitor under test are obtained; Based on the feature markers, the steady-state transition range of the capacitor under test is vectorized to obtain the voltage data packet of the capacitor under test. Extreme value detection is performed on the voltage data packet to obtain the initial voltage value of the capacitor under test.

[0009] In a preferred embodiment, when the charge / discharge characteristic trajectory tracking module performs voltage trajectory tracking on the initial voltage value to obtain the charge / discharge characteristic curve of the capacitor under test, it is specifically used for: Noise spectrum analysis is performed on the initial voltage value to obtain a clean reference signal for the capacitor under test; Based on the pure reference signal, a dynamic sliding window is constructed for the capacitor under test; Based on the dynamic sliding window, the real-time voltage data stream of the capacitor under test is segmented to obtain the data segment of the capacitor under test. Trend term separation is performed on the data segment to obtain the transient change characteristics of the capacitor under test; Inflection point detection and feature point marking are performed on the transient change characteristics to obtain the discrete feature sequence of the capacitor under test; The discrete feature sequence is smoothly fitted to obtain the charge-discharge characteristic curve of the capacitor under test.

[0010] In a preferred embodiment, when the dielectric polarization segmented slope identification module performs segmented slope identification on the charge saturation region and discharge cutoff region of the charge-discharge characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test, it is specifically used for: The characteristic interval of the charge-discharge characteristic curve is located to obtain the interval segment of the capacitor under test. The interval segment includes the data segment of the charge saturation region and the data segment of the discharge cutoff region. Local weighted scatter smoothing is applied to the interval segment to obtain the continuous trend baseline of the capacitor under test; By performing trend gradient analysis on the continuous trend baseline, the instantaneous rate of change of the interval segment is obtained; Segmented cluster analysis is performed on the instantaneous rate of change sequence to obtain the critical nodes of abrupt change in the capacitor under test; Based on the critical node, the continuous trend baseline is segmented and its features are analyzed to obtain the linearity characteristics and slope extreme values ​​of the capacitor under test. The linearity feature and the slope extremum are fused to obtain the internal dielectric polarization feature of the capacitor under test.

[0011] In a preferred embodiment, when the multidimensional fault feature weighted fusion module performs feature weighted fusion of the internal dielectric polarization features and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test, it is specifically used for: Ripple separation is performed on the environmental response signal to obtain the ripple component of the environmental response signal; The internal dielectric polarization characteristics and the ripple components are aligned to obtain the standardized characteristics of the capacitor under test. Correlation analysis is performed on the standardized features to obtain the correlation strength mapping relationship between the standardized features; Based on the correlation strength mapping relationship, determine the weighting coefficients of the standardized features; Based on the weighting coefficients, the standardized features are weighted and fused to obtain the fused feature values ​​of the standardized features; The fusion feature values ​​are subjected to dimensionality reduction extraction to obtain the multidimensional fault diagnosis vector of the capacitor under test.

[0012] In a preferred embodiment, the formula for calculating the fusion feature value is as follows: ; in, The fusion feature value, For the first The standardized value of the internal medium polarization characteristic, For the first The weighting coefficients of the internal medium polarization characteristics, For the first The standardized eigenvalues ​​of the ripple components, For the first The weighting coefficients of the ripple components, The total number of the internal dielectric polarization features. This represents the total number of ripple components.

[0013] In a preferred embodiment, when the closed-loop detection interactive feedback execution module performs interactive feedback on the multi-dimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test, it is specifically used for: The feature dimensions of the multidimensional fault diagnosis vector are normalized to obtain the standardized data sequence of the capacitor under test. Based on a preset fault knowledge base, the standardized data sequence is matched and analyzed to obtain the initial fault determination of the capacitor under test. The initial fault determination is corrected hierarchically to obtain multi-level feedback suggestions for the capacitor under test; The multi-level feedback suggestions are mapped to execution instructions for the capacitor under test to achieve closed-loop detection of the capacitor under test.

[0014] To address the above problems, the present invention also provides a multifunctional detection method for power capacitors, the method comprising: Step a: Obtain the access status monitoring signal of the capacitor under test, and perform baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test; Step b: Based on the basic electrical test environment, apply electrothermal stress superposition excitation to the capacitor under test to obtain the environmental response signal of the capacitor under test; Step c: Switch the constant current source of the environmental response signal to obtain the initial voltage value of the capacitor under test, and perform voltage trajectory tracking on the initial voltage value to obtain the charge and discharge characteristic curve of the capacitor under test; Step d: Identify the segmented slopes of the charging saturation region and the discharging cutoff region in the charge-discharge characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test. Step e: Perform feature weighted fusion of the internal dielectric polarization characteristics and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test; Step f: Perform interactive feedback on the multidimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention, through the synergistic processing of access state baseline calibration and electrothermal stress superposition excitation, can accurately construct a stable and reliable electrical testing environment, effectively filter out signal interference and improve the compatibility between the excitation signal and the capacitor under test, ensure the purity and accuracy of the test data, and at the same time achieve stable and controllable testing process.

[0016] 2. This invention utilizes charging and discharging characteristic trajectory tracking, dielectric polarization segment slope identification, and multi-dimensional fault feature weighted fusion technology to quickly extract core fault features inside the capacitor, simplifying the feature analysis process. Combined with a closed-loop detection interactive feedback execution mechanism, it can significantly improve fault diagnosis efficiency and the accuracy of detection results, achieving full-process intelligent and efficient power capacitor detection. Attached Figure Description

[0017] Figure 1 This is a system architecture diagram of a multifunctional detection system for power capacitors provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating a multifunctional detection method for power capacitors according to an embodiment of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments belong to some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “said” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0021] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0022] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0023] In practice, the server-side equipment deployed in a multi-functional power capacitor testing system may consist of one or more devices. This multi-functional power capacitor testing system can be implemented as a business instance, a virtual machine, or hardware devices. For example, it can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, it can be understood as software deployed on a cloud node, providing a multi-functional power capacitor testing system to various user terminals. Alternatively, it can be implemented as a virtual machine deployed on one or more devices in a cloud node, with application software installed to manage various user terminals. Or, it can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide a multi-functional power capacitor testing system to various user terminals.

[0024] In terms of implementation, the multifunctional power capacitor detection system and the user terminal are mutually compatible. That is, if the multifunctional power capacitor detection system is implemented as an application installed on a cloud service platform, the user terminal is implemented as a client that establishes a communication connection with the application; or if the multifunctional power capacitor detection system is implemented as a website, the user terminal is implemented as a webpage; or if the multifunctional power capacitor detection system is implemented as a cloud service platform, the user terminal is implemented as a mini-program in an instant messaging application.

[0025] like Figure 1 The figure shown is a system architecture diagram of a multifunctional detection system for power capacitors provided in an embodiment of the present invention.

[0026] The multifunctional power capacitor testing system 100 described in this invention can be installed on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed in the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed into a website. Depending on the functions implemented, the multifunctional power capacitor testing system 100 may include an access status baseline calibration module 101, an electrothermal stress superposition excitation module 102, a charge / discharge characteristic trajectory tracking module 103, a dielectric polarization segment slope identification module 104, a multidimensional fault feature weighted fusion module 105, and a closed-loop detection interactive feedback execution module 106. The module described in this invention can also be called a unit, referring to a series of computer program segments that can be executed by an electronic device's processor and perform a fixed function, stored in the electronic device's memory.

[0027] In this embodiment of the invention, in a multifunctional power capacitor testing system, each of the above-mentioned modules can be implemented independently and called upon other modules. This "calling" can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. The multifunctional power capacitor testing system provided by this embodiment of the invention allows for adjustment of the system's applicability by adding modules and directly calling them, without modifying the program code. This enables cluster-based horizontal expansion, facilitating quick and flexible expansion of the system. In practical applications, these modules can be located in the same or different devices, or in virtual devices, such as service instances on a cloud server.

[0028] The following describes, with reference to specific embodiments, each component and its specific workflow of a multifunctional power capacitor testing system: The access status baseline calibration module 101 acquires the access status monitoring signal of the capacitor under test and performs baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test. In this embodiment of the invention, when the access status baseline calibration module acquires the access status monitoring signal of the capacitor under test and performs baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test, it is specifically used for: Acquire the connection status monitoring signal of the capacitor under test; The DC component of the access status monitoring signal is filtered out to obtain the DC-free signal of the capacitor under test; The frequency domain characteristic spectrum of the capacitor under test is obtained by performing spectral analysis on the DC component signal. Based on the energy distribution density in the frequency domain characteristic spectrum, the strong interference frequency band of the capacitor under test is identified to generate the adaptive notch filter parameters of the capacitor under test. Based on the adaptive notch filter parameters, the DC component removal signal is iteratively filtered to obtain the pure timing signal of the capacitor under test. The steady-state reference range of the capacitor under test is obtained by performing sliding window variance statistics on the pure time series signal. Based on the steady-state reference range, zero-point drift compensation is performed on the pure timing signal to obtain the basic electrical test environment of the capacitor under test.

[0029] The voltage and current acquisition unit continuously and uninterruptedly acquires data from the test circuit at a preset high sampling frequency, fully recording the voltage and current timing data of the capacitor under test from the moment of connection to the test circuit until stable operation. Simultaneously, the environmental electrical noise data of the test circuit under no-load condition is acquired as a reference benchmark, comprehensively covering the transient and steady-state electrical characteristics of the capacitor under test during the connection process, and generating the connection status monitoring signal of the capacitor under test.

[0030] The collected access status monitoring signals are processed point by point. The DC and AC components in the signal are accurately separated by the preset filter cutoff frequency. While filtering out the DC component, the signal phase distortion and amplitude distortion generated during the filtering process are suppressed to the greatest extent, and the effective electrical characteristic information in the AC component is completely preserved, so as to obtain the DC component-free signal of the capacitor under test.

[0031] The DC component-free signal is processed by full-band fast Fourier transform, which converts the continuous voltage and current signal in the time domain into discrete frequency components in the frequency domain. The amplitude and phase information corresponding to each frequency point are extracted, and a complete three-dimensional correspondence between frequency, amplitude and phase is constructed to fully present the energy distribution of the signal in the full frequency band and generate the frequency domain characteristic spectrum of the capacitor under test.

[0032] The energy distribution of all frequency points in the frequency domain feature spectrum is traversed. The entire frequency band is divided into several continuous sub-bands with a preset fixed frequency interval. The cumulative energy value in each sub-band is counted segment by segment. The energy proportion of all sub-bands is compared to identify strong interference bands with a significantly higher energy proportion than other sub-bands. Based on the center frequency, bandwidth range and energy peak value of the strong interference band, adaptive notch filter parameters adapted to the interference band are generated.

[0033] An infinite impulse response notch filter structure is constructed based on adaptive notch parameters for the corresponding frequency band. Precise filter coefficients are designed for the frequency characteristics of the strong interference band. The DC component signal is subjected to multiple rounds of iterative filtering. After each round of filtering, the signal spectrum is re-analyzed to detect the energy residue in the strong interference band. The signal energy of the interference band is gradually attenuated until the energy ratio of the interference band drops below the preset threshold and there is no obvious interference component residue, thus obtaining the pure timing signal of the capacitor under test.

[0034] The clean time-series signal is continuously segmented by a sliding window of fixed time length. The window slides along the time-series direction successively with a preset step size to ensure that no data is missed between adjacent windows. The variance of the time-series signal in each window segment is calculated, and the distribution range of the variance of all windows is statistically analyzed. The variance data of windows with abnormal fluctuations are removed by using the Grubbs criterion. Based on the variance statistics of the remaining effective windows, the stable interval of signal fluctuation is determined, and the steady-state reference interval of the capacitor under test is generated.

[0035] Using the center value of the steady-state reference range as a reference, the amplitude of the pure timing signal is corrected point by point. For the zero-point offset caused by circuit temperature drift, device aging, and environmental electromagnetic interference, corresponding compensation is made according to the magnitude of the deviation from the reference. The amplitude of all timing points is uniformly calibrated to the steady-state reference range, completely eliminating the zero-point drift phenomenon of the signal and obtaining the basic electrical test environment of the capacitor under test.

[0036] The beneficial effects are as follows: Through a closed-loop process involving DC component filtering, spectrum analysis, adaptive notch filtering, sliding window variance statistics, and zero-point drift compensation, the DC component, strong interference noise, and zero-point drift problem in the condition monitoring signal of the capacitor under test can be completely eliminated. This generates a stable, clean, fluctuating, and dimensionally consistent basic electrical test environment, providing a highly reliable signal foundation for subsequent electrothermal stress excitation, feature extraction, and fault diagnosis of power capacitors. This significantly improves the accuracy and stability of electrical testing, ensures the precision and reliability of power capacitor condition detection results, and adapts to complex interference environments under different test scenarios, enhancing the environmental adaptability and robustness of the detection system and meeting the high-precision requirements of power capacitor condition detection throughout its entire life cycle.

[0037] The electrothermal stress superposition excitation module 102, based on the basic electrical test environment, performs electrothermal stress superposition excitation on the capacitor under test to obtain the environmental response signal of the capacitor under test; In this embodiment of the invention, when the electrothermal stress superposition excitation module performs electrothermal stress superposition excitation on the capacitor under test based on the basic electrical test environment to obtain the environmental response signal of the capacitor under test, it is specifically used for: The background impedance data in the basic electrical test environment is processed by time-frequency domain transformation to obtain the load characteristic vector of the capacitor under test; Based on a preset multi-dimensional excitation strategy library, the load characteristic vector is indexed and matched to obtain the excitation parameters of the capacitor under test. The dynamic excitation parameters are time-series segmented and recombined to obtain the composite excitation sequence of the capacitor under test; Based on the composite excitation sequence, the capacitor under test is subjected to segmented data-driven excitation to obtain the original response data stream of the capacitor under test; The original response data stream is subjected to sliding window filtering to obtain the environmental response signal of the capacitor under test.

[0038] The background impedance data contained in the basic electrical test environment is subjected to complete time-frequency domain transformation. First, the continuous change data of the background impedance in the time domain is discretized and sampled at fixed time intervals. Then, the sampled time-domain impedance data is converted into impedance distribution information in the frequency domain. The impedance amplitude, phase and change trend corresponding to different frequency points are extracted respectively. The time-domain dynamic features and frequency-domain static features are integrated and collected to form a feature set that can comprehensively characterize the external load state of the capacitor under test, and the load characteristic vector of the capacitor under test is obtained.

[0039] All pre-defined excitation strategy entries are retrieved from the pre-defined multi-dimensional excitation strategy library. Each entry contains complete information such as voltage excitation level, temperature excitation range, current excitation waveform, and duration. Each feature in the load characteristic vector is precisely compared with the corresponding entry in the multi-dimensional excitation strategy library dimension by dimension. The indexing and scheme locking are completed according to the principle of highest feature matching degree and best adaptability. The excitation configuration content that perfectly matches the current load state and test environment of the capacitor under test is determined, and the excitation parameters of the capacitor under test are obtained.

[0040] According to the preset time progression rules and excitation superposition logic, the excitation parameters obtained by index matching are segmented and orderly recombined. Different types of parameters such as voltage excitation, thermal excitation, and current excitation are classified and arranged according to their action stages, forming a structure in time sequence where the preceding and subsequent excitations are smoothly connected. This allows different excitation parameters to form a continuous and controllable combination on the time axis, resulting in the composite excitation sequence of the capacitor under test.

[0041] Strictly following the segmented execution requirements set by the composite excitation sequence, electrical and thermal stresses with corresponding parameters are applied sequentially to the capacitor under test. The electrical and thermal excitations are applied to the capacitor under test in a synchronous superposition manner. During each excitation application process, the port voltage, loop current, surface temperature, and internal response data of the capacitor under test are collected in real time. The complete correspondence between the excitation input and the device output is fully preserved, forming a continuous and uninterrupted set of response data, and the original response data stream of the capacitor under test is obtained.

[0042] A sliding window of fixed time length is used to slide continuously along the time sequence of the original response data stream. Each time the window slides, all data in the current window is processed. First, abnormal data points that deviate from the normal range are removed. Then, the mean smoothing calculation is performed on the remaining valid data to preserve the true trend of the data and completely filter out random fluctuations and high-frequency noise. This completes the smoothing and noise reduction processing of the entire data stream, and the environmental response signal of the capacitor under test is obtained.

[0043] The beneficial effects are as follows: the load characteristic vector is accurately extracted by the time-frequency domain transformation of the background impedance, and the excitation parameters are matched with a high degree of adaptability by combining a multi-dimensional excitation strategy library. After time-series segmentation and recombination, a stable and controllable composite excitation sequence is formed. Then, the actual state of the equipment is excited by the segmented electrothermal stress superposition. Finally, the original response data is smoothed and denoised by sliding window filtering. It can efficiently obtain pure, stable and highly recognizable environmental response signals, providing a reliable data foundation for subsequent charge and discharge characteristic trajectory tracking, dielectric polarization feature identification and fault diagnosis, and significantly improving the adaptability, stability and accuracy of the detection process.

[0044] The charge / discharge characteristic trajectory tracking module 103 switches the constant current source of the environmental response signal to obtain the initial voltage value of the capacitor under test, and performs voltage trajectory tracking on the initial voltage value to obtain the charge / discharge characteristic curve of the capacitor under test. In this embodiment of the invention, the charge / discharge characteristic trajectory tracking module performs constant current source switching on the environmental response signal to obtain the initial voltage value of the capacitor under test, specifically for: The environmental response signal is time-series segmented to obtain a multidimensional data segment of the capacitor under test. By performing correlation feature matching on the multidimensional data fragments, the feature marker points of the capacitor under test are obtained; Based on the feature markers, the steady-state transition range of the capacitor under test is vectorized to obtain the voltage data packet of the capacitor under test. Extreme value detection is performed on the voltage data packet to obtain the initial voltage value of the capacitor under test.

[0045] When the charge / discharge characteristic trajectory tracking module performs voltage trajectory tracking on the initial voltage value to obtain the charge / discharge characteristic curve of the capacitor under test, it is specifically used for: Noise spectrum analysis is performed on the initial voltage value to obtain a clean reference signal for the capacitor under test; Based on the pure reference signal, a dynamic sliding window is constructed for the capacitor under test; Based on the dynamic sliding window, the real-time voltage data stream of the capacitor under test is segmented to obtain the data segment of the capacitor under test. Trend term separation is performed on the data segment to obtain the transient change characteristics of the capacitor under test; Inflection point detection and feature point marking are performed on the transient change characteristics to obtain the discrete feature sequence of the capacitor under test; The discrete feature sequence is smoothly fitted to obtain the charge-discharge characteristic curve of the capacitor under test.

[0046] The environmental response signal is continuously and uniformly segmented according to a preset fixed time step, and the originally continuous and uninterrupted environmental response signal is decomposed into multiple data units of the same length, which are interconnected and have no data omissions. Each data unit completely retains multi-dimensional electrical change information such as voltage, current, and temperature, ensuring that the entire process of the capacitor under test from the start of the excitation response to the state tending to stabilize is covered, thus obtaining multi-dimensional data segments of the capacitor under test.

[0047] The voltage, current, and temperature dimensions within the multidimensional data segment are synchronously correlated and compared. The changing trends and numerical correspondences of different dimensions at the same time node are analyzed. Consistent features that can reflect state switching are selected. Based on these correlation features, precise positioning and marking are completed at the time sequence position, locking the most representative key position of signal change, and obtaining the feature marking points of the capacitor under test.

[0048] Using feature markers as interval boundaries, the complete interval of the capacitor under test transitioning from the unsteady state at the initial stage of excitation to the steady state required for testing is accurately defined. All voltage-related data within this steady-state transition interval are uniformly organized and arranged in chronological order, transforming the originally scattered and disordered discrete data into an ordered vector form with direction and magnitude, thus fully preserving the details of voltage changes within the interval and obtaining the voltage data packet of the capacitor under test.

[0049] The voltage data packet is traversed point by point and compared horizontally to completely scan the size of each voltage data in the data packet. The extreme points in the voltage data sequence are accurately identified. The extreme points correspond to the initial voltage state of the capacitor under test at the moment of constant current source switching, and the initial voltage value of the capacitor under test is obtained.

[0050] The original signal segment containing the initial voltage value is subjected to full-band noise spectrum decomposition, which completely separates the effective electrical components from various noise components in the signal, accurately locates and removes high-frequency random noise and low-frequency interference components, and completely retains the reference information that can represent the real voltage change, thus obtaining the pure reference signal of the capacitor under test.

[0051] Based on the amplitude range, rate of change, and timing length of the pure reference signal, the initial width, sliding step size, and adaptive adjustment rules of the sliding window are determined, so that the window size can automatically adapt to the speed of change of the voltage signal, thus constructing a dynamic sliding window that perfectly matches the voltage change characteristics of the capacitor under test.

[0052] Based on the constructed dynamic sliding window, the real-time voltage data stream generated by the capacitor under test during charging and discharging is continuously segmented. The window slides smoothly along the time sequence direction, and each segment ensures that the complete local features of voltage change are covered. There is no data overlap or data loss between adjacent windows, thus obtaining the data segment of the capacitor under test.

[0053] For each segment of data obtained, the trend term and fluctuation term are separated. The long-term trend representing the overall direction of change in the data is completely separated from the short-term fluctuation representing instantaneous change. Irrelevant stationary components are removed, and the rapid change information of voltage during charging and discharging is accurately extracted to obtain the transient change characteristics of the capacitor under test.

[0054] The transient change characteristics are fully traversed to detect the locations where the signal change direction and rate change abruptly occur, and all inflection points are accurately located. These inflection points are then marked as feature points in chronological order to form an ordered set of feature points, thus obtaining the discrete feature sequence of the capacitor under test.

[0055] All feature points in the discrete feature sequence are subjected to continuous smooth fitting. The discrete feature points are connected into a continuous and smooth curve by curve fitting, which completely restores the voltage change trajectory of the capacitor under test during the entire charging and discharging process, and obtains the charging and discharging characteristic curve of the capacitor under test.

[0056] The beneficial effects are as follows: by accurately obtaining the initial voltage value through time segmentation, correlation feature matching, data vectorization and extreme value detection, and then through noise spectrum analysis, dynamic sliding window construction, data segmentation and extraction, trend term separation, inflection point detection and smooth fitting, the charging and discharging trajectory can be completely restored. This can accurately capture the real voltage changes during the charging and discharging process of the capacitor under test, effectively filter out noise interference and abnormal fluctuations, and generate high-definition and highly continuous charging and discharging characteristic curves. This provides an accurate and reliable basis for subsequent identification of dielectric polarization characteristics and greatly improves the accuracy and stability of fault feature extraction.

[0057] The dielectric polarization segmented slope identification module 104 identifies the segmented slope of the charging saturation region and the discharging cutoff region in the charging and discharging characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test. In this embodiment of the invention, when the dielectric polarization segmented slope identification module performs segmented slope identification on the charge saturation region and discharge cutoff region of the charge-discharge characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test, it is specifically used for: The characteristic interval of the charge-discharge characteristic curve is located to obtain the interval segment of the capacitor under test. The interval segment includes the data segment of the charge saturation region and the data segment of the discharge cutoff region. Local weighted scatter smoothing is applied to the interval segment to obtain the continuous trend baseline of the capacitor under test; By performing trend gradient analysis on the continuous trend baseline, the instantaneous rate of change of the interval segment is obtained; Segmented cluster analysis is performed on the instantaneous rate of change sequence to obtain the critical nodes of abrupt change in the capacitor under test; Based on the critical node, the continuous trend baseline is segmented and its features are analyzed to obtain the linearity characteristics and slope extreme values ​​of the capacitor under test. The linearity feature and the slope extremum are fused to obtain the internal dielectric polarization feature of the capacitor under test.

[0058] The charge-discharge characteristic curve is scanned across the entire domain according to the rate of voltage change and the degree of numerical stability. The charging saturation region is located based on the numerical range where the curve tends to be stable during the charging phase, and the discharge cutoff region is located based on the numerical range where the curve tends to terminate during the discharging phase. All data points in the two regions are extracted and the temporal integrity is maintained to obtain the charging saturation region data segment and the discharge cutoff region data segment of the capacitor under test.

[0059] Local weighted scatter smoothing is performed on the data segments of the charging saturation region and the discharging cutoff region respectively. A fixed number of neighboring data points are selected as the center of each data point to form a local calculation interval. Data within the interval are assigned distance-related weights and weighted calculations are performed. The original data is replaced point by point to eliminate discrete fluctuations and glitch interference, and the scattered data points are converted into continuous and stable changing lines to obtain the continuous trend baseline of the capacitor under test.

[0060] For each data point on the continuous trend baseline, the numerical difference and time difference between adjacent points are calculated. The change in the baseline value per unit time is used to characterize the rate of change at the current position. After completing the calculation point by point, a complete change rate sequence is formed, which fully reflects the intensity of change of the baseline at different positions, and the instantaneous change rate of the interval segment is obtained.

[0061] Segmented clustering analysis was performed on the instantaneous rate of change sequence according to its numerical value and change pattern. Instantaneous rates of change with similar values ​​and consistent trends were classified into the same category. The positions where there were obvious numerical jumps between categories were identified. These positions were the locations where the curve state underwent substantial changes, and the abrupt critical nodes of the capacitor under test were obtained.

[0062] Using the critical node of abrupt change as the dividing point, the continuous trend baseline is segmented and analyzed. The deviation of each segment from the ideal straight line is calculated to characterize the linearity. At the same time, the maximum and minimum values ​​of the instantaneous rate of change within each segment are found to determine the limit of the change range of each segment. Stable feature information is extracted to obtain the linearity characteristics and slope extreme values ​​of the capacitor under test.

[0063] The linearity characteristics and slope extrema are merged point by point according to the time sequence correspondence. The information such as the change law, stability degree and limit amplitude contained in the two types of characteristics are integrated into a unified feature set, which fully reflects the polarization behavior and state change of the internal dielectric of the capacitor under test during the charging and discharging process, and obtains the internal dielectric polarization characteristics of the capacitor under test.

[0064] The beneficial effects are that by accurately locating the key charging and discharging intervals, eliminating interference through local weighted smoothing, quantifying changes through trend gradient analysis, locking abrupt change nodes through cluster analysis, extracting linearity and slope extrema in segments, and completing feature fusion, it can accurately extract the polarization change law of the internal medium of the capacitor under test, effectively filter curve noise and non-critical fluctuations, and generate internal medium polarization characteristics that can truly reflect the aging and defect state of the medium, providing core and reliable feature basis for subsequent multidimensional fault diagnosis.

[0065] The multidimensional fault feature weighted fusion module 105 performs feature weighted fusion of the internal dielectric polarization features and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test. In this embodiment of the invention, when the multidimensional fault feature weighted fusion module performs feature weighted fusion of the internal dielectric polarization features and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test, it is specifically used for: Ripple separation is performed on the environmental response signal to obtain the ripple component of the environmental response signal; The internal dielectric polarization characteristics and the ripple components are aligned to obtain the standardized characteristics of the capacitor under test. Correlation analysis is performed on the standardized features to obtain the correlation strength mapping relationship between the standardized features; Based on the correlation strength mapping relationship, determine the weighting coefficients of the standardized features; Based on the weighting coefficients, the standardized features are weighted and fused to obtain the fused feature values ​​of the standardized features; The fusion feature values ​​are subjected to dimensionality reduction extraction to obtain the multidimensional fault diagnosis vector of the capacitor under test.

[0066] The formula for calculating the fusion feature value is as follows: ; in, The fusion feature value, For the first The standardized value of the internal medium polarization characteristic, For the first The weighting coefficients of the internal medium polarization characteristics, For the first The standardized eigenvalues ​​of the ripple components, For the first The weighting coefficients of the ripple components, The total number of the internal dielectric polarization features. This represents the total number of ripple components.

[0067] The environmental response signal is decomposed and its components are extracted across the entire frequency band. Through signal separation processing, the fundamental stable components and high-frequency fluctuation components in the environmental response signal are distinguished. The periodic and non-periodic fluctuation components superimposed on the stable signal are accurately extracted, and the amplitude, frequency and phase information of the ripple are completely preserved to obtain the ripple component of the environmental response signal.

[0068] The internal dielectric polarization characteristics and ripple components are normalized according to a unified time axis and numerical range. The characteristic data from two different sources, with different dimensions and different numerical ranges are adjusted to the same time reference and numerical scale to ensure that each set of characteristic data is completely matched in terms of time nodes and numerical distribution. This eliminates the time misalignment and magnitude difference between characteristics and obtains the standardized characteristics of the capacitor under test.

[0069] The correlation between the standardized internal medium polarization characteristics and ripple components is calculated dimension by dimension. The correlation between the numerical changes of each polarization characteristic and each ripple characteristic is comprehensively analyzed. A complete correspondence matrix is ​​constructed through the synchronous change law between the characteristics, which clearly presents the degree of influence and dependence between different characteristics, and obtains the correlation strength mapping relationship between standardized characteristics.

[0070] Based on the importance and mutual influence weights of the features presented in the correlation strength mapping relationship, a corresponding contribution ratio is assigned to each standardized feature. Features with high contribution to fault diagnosis are assigned a higher ratio, and features with low contribution are assigned a corresponding ratio, so that the weight allocation is fully consistent with the actual representation ability of the features to the equipment status, and the weighting coefficient of the standardized features is determined.

[0071] Based on the determined weighting coefficients, each standardized feature is numerically combined with its corresponding weighting coefficient. The weighting results of all features are summarized and integrated, so that different features participate in the overall calculation according to their own importance, forming a unified numerical result that can comprehensively reflect the multi-dimensional status of the equipment, and obtaining the fused feature value of the standardized features.

[0072] The fusion feature value is obtained by summing the products of the standardized values ​​of all internal dielectric polarization features and their corresponding weighting coefficients, and then summing the products of the standardized values ​​of all ripple components and their corresponding weighting coefficients. The two summation results are then obtained, achieving a weighted fusion of the two types of features: internal dielectric polarization features and ripple components. This generates a unified fusion feature value, which is used for subsequent power capacitor fault diagnosis and condition assessment. The standardization process eliminates the dimensional differences between different features, and the determination of the weighting coefficients ensures that the contribution of each feature to fault diagnosis is matched, ensuring that the fusion feature value can accurately characterize the internal state of the power capacitor.

[0073] The standardized values ​​of the internal dielectric polarization characteristics are generated by standardizing the internal dielectric polarization characteristics extracted in the power capacitor charging and discharging characteristic trajectory tracking and dielectric polarization segment slope identification steps. The weighting coefficients corresponding to the internal dielectric polarization characteristics are determined by constructing a judgment matrix, calculating a weight vector, and completing a consistency check based on the contribution of each internal dielectric polarization characteristic to fault diagnosis. The total number of internal dielectric polarization characteristics represents the statistical number of effective internal dielectric polarization characteristics extracted in this step. The standardized values ​​of the ripple components are generated by standardizing the ripple components extracted in the power capacitor electrothermal stress excitation response signal acquisition and preprocessing steps. The weighting coefficients corresponding to the ripple components are determined by constructing a judgment matrix, calculating a weight vector, and completing a consistency check based on the contribution of each ripple component to fault diagnosis. The total number of ripple components represents the statistical number of effective ripple components extracted in this step. The standardization process uses the z-score standardization method to map the original values ​​of the two types of characteristics to the same dimension range, eliminating the difference between dimensions and numerical magnitudes and ensuring parameter dimension uniformity.

[0074] The fusion characteristic value increases synchronously with the increase of the normalized value of the internal medium polarization characteristic, synchronously with the increase of the weighting coefficient of the internal medium polarization characteristic, synchronously with the increase of the normalized value of the ripple component, synchronously with the increase of the weighting coefficient of the ripple component, synchronously with the increase of the total number of internal medium polarization characteristics, and synchronously with the increase of the total number of ripple components.

[0075] The dimensionality of the fused feature values ​​is reduced by removing redundant and repetitive information components, while retaining the core information that is most distinguishable and representative for fault diagnosis. The high-dimensional fused features are compressed into a low-dimensional and highly condensed feature set, which can be directly used for fault determination and status assessment, resulting in a multi-dimensional fault diagnosis vector for the capacitor under test.

[0076] The beneficial effect is that through the complete process of ripple separation, data alignment, correlation analysis, weighting coefficient determination, weighted fusion calculation and dimensionality reduction extraction, it can efficiently integrate the two core features of internal medium polarization characteristics and ripple components, eliminate the differences and redundancy between features, highlight key fault information, and generate multi-dimensional fault diagnosis vectors with simplified dimensions, strong representation ability and high recognition, providing accurate and reliable core basis for subsequent closed-loop detection and fault judgment.

[0077] The closed-loop detection interactive feedback execution module 106 provides interactive feedback to the multidimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test.

[0078] In this embodiment of the invention, when the closed-loop detection interactive feedback execution module performs interactive feedback on the multi-dimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test, it is specifically used for: The feature dimensions of the multidimensional fault diagnosis vector are normalized to obtain the standardized data sequence of the capacitor under test. Based on a preset fault knowledge base, the standardized data sequence is matched and analyzed to obtain the initial fault determination of the capacitor under test. The initial fault determination is corrected hierarchically to obtain multi-level feedback suggestions for the capacitor under test; The multi-level feedback suggestions are mapped to execution instructions for the capacitor under test to achieve closed-loop detection of the capacitor under test.

[0079] The values ​​of all feature dimensions in the multidimensional fault diagnosis vector are uniformly scaled and shifted to map all feature data into the same numerical range, eliminating the differences in numerical magnitude and distribution deviation between different feature dimensions, so that each feature has an equal basis for comparison and judgment, forming a continuous, regular and uniform data arrangement, and obtaining a standardized data sequence of the capacitor under test.

[0080] The system retrieves all standard entries from the pre-set fault knowledge base, which include various typical fault modes, feature distribution ranges, and status judgment rules. It then compares and calculates the similarity between each feature of the standardized data sequence and the corresponding standard entry in the knowledge base. Based on the principle of the highest feature matching degree, it locks the most suitable fault type and status level, completes the preliminary equipment status judgment, and obtains the initial fault judgment of the capacitor under test.

[0081] Following a progressive rule of detection accuracy level, equipment importance level, and fault risk level, the initial fault judgment results are verified and adjusted in multiple rounds. First, deviations are corrected based on real-time detection data, then the judgment conclusions are optimized by combining historical equipment operation data, and finally the final calibration is completed based on the on-site test environment parameters. This process gradually improves the accuracy and reliability of the judgment results, forms guidance content covering different application scenarios, and obtains multi-level feedback suggestions for the capacitor under test.

[0082] The multi-level feedback suggestions, including status assessment, fault location, handling strategies, and parameter adjustment, are converted into control and operation commands that the test system can directly recognize and execute. These commands can directly drive the excitation module, acquisition module, and analysis module to complete operations such as parameter adjustment, repeated testing, and data verification, forming a complete closed-loop process from diagnosis to execution to verification, thus realizing fully automated closed-loop testing of the capacitor under test.

[0083] The beneficial effects are that through the closed-loop execution process of feature dimension normalization, fault knowledge base matching, multi-level judgment correction and instruction mapping, multi-dimensional fault diagnosis vectors can be quickly transformed into accurate and reliable fault judgment results and executable operation instructions, realizing fully automatic closed-loop operation of detection, analysis, feedback and correction, improving the real-time performance and accuracy of fault diagnosis, and ensuring efficient and stable operation of the entire power capacitor detection process.

[0084] Reference Figure 2 The diagram shown is a flowchart illustrating a multifunctional detection method for power capacitors according to an embodiment of the present invention. In this embodiment, the multifunctional detection method for power capacitors includes: Step a: Obtain the access status monitoring signal of the capacitor under test, and perform baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test; Step b: Based on the basic electrical test environment, apply electrothermal stress superposition excitation to the capacitor under test to obtain the environmental response signal of the capacitor under test; Step c: Switch the constant current source of the environmental response signal to obtain the initial voltage value of the capacitor under test, and perform voltage trajectory tracking on the initial voltage value to obtain the charge and discharge characteristic curve of the capacitor under test; Step d: Identify the segmented slopes of the charging saturation region and the discharging cutoff region in the charge-discharge characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test. Step e: Perform feature weighted fusion of the internal dielectric polarization characteristics and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test; Step f: Perform interactive feedback on the multidimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test.

[0085] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0086] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multifunctional testing system for power capacitors, characterized in that, The system includes an access state baseline calibration module, an electrothermal stress superposition excitation module, a charge / discharge characteristic trajectory tracking module, a dielectric polarization segmented slope identification module, a multi-dimensional fault feature weighted fusion module, and a closed-loop detection interactive feedback execution module, wherein: The access status baseline calibration module acquires the access status monitoring signal of the capacitor under test and performs baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test. The electrothermal stress superposition excitation module, based on the basic electrical test environment, performs electrothermal stress superposition excitation on the capacitor under test to obtain the environmental response signal of the capacitor under test; The charge / discharge characteristic trajectory tracking module switches the constant current source of the environmental response signal to obtain the initial voltage value of the capacitor under test, and performs voltage trajectory tracking on the initial voltage value to obtain the charge / discharge characteristic curve of the capacitor under test. The dielectric polarization segmented slope identification module identifies the segmented slope of the charging saturation region and the discharging cutoff region in the charging and discharging characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test. The multidimensional fault feature weighted fusion module performs feature weighted fusion of the internal dielectric polarization features and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test. The closed-loop detection interactive feedback execution module provides interactive feedback to the multi-dimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test.

2. The multifunctional testing system for power capacitors as described in claim 1, characterized in that, When the access status baseline calibration module acquires the access status monitoring signal of the capacitor under test and performs baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test, it is specifically used for: Acquire the connection status monitoring signal of the capacitor under test; The DC component of the access status monitoring signal is filtered out to obtain the DC-free signal of the capacitor under test; The frequency domain characteristic spectrum of the capacitor under test is obtained by performing spectral analysis on the DC component signal. Based on the energy distribution density in the frequency domain characteristic spectrum, the strong interference frequency band of the capacitor under test is identified, so as to generate the adaptive notch filter parameters of the capacitor under test. Based on the adaptive notch filter parameters, the DC component removal signal is subjected to iterative filtering to obtain the pure timing signal of the capacitor under test. The steady-state reference range of the capacitor under test is obtained by performing sliding window variance statistics on the pure time series signal. Based on the steady-state reference range, zero-point drift compensation is performed on the pure timing signal to obtain the basic electrical test environment of the capacitor under test.

3. The multifunctional testing system for power capacitors as described in claim 1, characterized in that, When the electrothermal stress superposition excitation module performs electrothermal stress superposition excitation on the capacitor under test based on the basic electrical test environment to obtain the environmental response signal of the capacitor under test, it is specifically used for: The background impedance data in the basic electrical test environment is processed by time-frequency domain transformation to obtain the load characteristic vector of the capacitor under test; Based on a preset multi-dimensional excitation strategy library, the load characteristic vector is indexed and matched to obtain the excitation parameters of the capacitor under test. The dynamic excitation parameters are time-series segmented and recombined to obtain the composite excitation sequence of the capacitor under test; Based on the composite excitation sequence, the capacitor under test is subjected to segmented data-driven excitation to obtain the original response data stream of the capacitor under test; The original response data stream is subjected to sliding window filtering to obtain the environmental response signal of the capacitor under test.

4. The multifunctional testing system for power capacitors as described in claim 1, characterized in that, The charge / discharge characteristic trajectory tracking module performs constant current source switching on the environmental response signal to obtain the initial voltage value of the capacitor under test, specifically for: The environmental response signal is time-series segmented to obtain a multidimensional data segment of the capacitor under test. By performing correlation feature matching on the multidimensional data fragments, the feature marker points of the capacitor under test are obtained; Based on the feature markers, the steady-state transition range of the capacitor under test is vectorized to obtain the voltage data packet of the capacitor under test. Extreme value detection is performed on the voltage data packet to obtain the initial voltage value of the capacitor under test.

5. The multifunctional testing system for power capacitors as described in claim 1, characterized in that, When the charge / discharge characteristic trajectory tracking module performs voltage trajectory tracking on the initial voltage value to obtain the charge / discharge characteristic curve of the capacitor under test, it is specifically used for: Noise spectrum analysis is performed on the initial voltage value to obtain a clean reference signal for the capacitor under test; Based on the pure reference signal, a dynamic sliding window is constructed for the capacitor under test; Based on the dynamic sliding window, the real-time voltage data stream of the capacitor under test is segmented to obtain the data segment of the capacitor under test. Trend term separation is performed on the data segment to obtain the transient change characteristics of the capacitor under test; Inflection point detection and feature point marking are performed on the transient change characteristics to obtain the discrete feature sequence of the capacitor under test; The discrete feature sequence is smoothly fitted to obtain the charge-discharge characteristic curve of the capacitor under test.

6. The multifunctional testing system for power capacitors as described in claim 1, characterized in that, When the dielectric polarization segmented slope identification module performs segmented slope identification on the charge saturation region and discharge cutoff region of the charge-discharge characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test, it is specifically used for: The characteristic interval of the charge-discharge characteristic curve is located to obtain the interval segment of the capacitor under test. The interval segment includes the data segment of the charge saturation region and the data segment of the discharge cutoff region. Local weighted scatter smoothing is applied to the interval segment to obtain the continuous trend baseline of the capacitor under test; By performing trend gradient analysis on the continuous trend baseline, the instantaneous rate of change of the interval segment is obtained; Segmented cluster analysis is performed on the instantaneous rate of change sequence to obtain the critical nodes of abrupt change in the capacitor under test; Based on the critical node, the continuous trend baseline is segmented and its features are analyzed to obtain the linearity characteristics and slope extreme values ​​of the capacitor under test. The linearity feature and the slope extremum are fused to obtain the internal dielectric polarization feature of the capacitor under test.

7. The multifunctional testing system for power capacitors as described in claim 1, characterized in that, When the multidimensional fault feature weighted fusion module performs feature weighted fusion of the internal dielectric polarization characteristics and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test, it is specifically used for: Ripple separation is performed on the environmental response signal to obtain the ripple component of the environmental response signal; The internal dielectric polarization characteristics and the ripple components are aligned to obtain the standardized characteristics of the capacitor under test. Correlation analysis is performed on the standardized features to obtain the correlation strength mapping relationship between the standardized features; Based on the correlation strength mapping relationship, determine the weighting coefficients of the standardized features; Based on the weighting coefficients, the standardized features are weighted and fused to obtain the fused feature values ​​of the standardized features; The fusion feature values ​​are subjected to dimensionality reduction extraction to obtain the multidimensional fault diagnosis vector of the capacitor under test. The formula for calculating the fusion feature value is as follows: ； in, The fusion feature value, For the first The standardized value of the internal medium polarization characteristic, For the first The weighting coefficients of the internal medium polarization characteristics, For the first The standardized eigenvalues ​​of the ripple components, For the first The weighting coefficients of the ripple components. The total number of the internal dielectric polarization features. This represents the total number of ripple components.

8. The multifunctional testing system for power capacitors as described in claim 1, characterized in that, When the closed-loop detection interactive feedback execution module performs interactive feedback on the multi-dimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test, it is specifically used for: The feature dimensions of the multidimensional fault diagnosis vector are normalized to obtain the standardized data sequence of the capacitor under test. Based on a preset fault knowledge base, the standardized data sequence is matched and analyzed to obtain the initial fault determination of the capacitor under test. The initial fault determination is corrected hierarchically to obtain multi-level feedback suggestions for the capacitor under test; The multi-level feedback suggestions are mapped to execution instructions for the capacitor under test to achieve closed-loop detection of the capacitor under test.

9. A multifunctional detection method for power capacitors, characterized in that, The method for using the multifunctional detection system for power capacitors according to claim 1: Step a: Obtain the access status monitoring signal of the capacitor under test, and perform baseline calibration on the access status monitoring signal to obtain the basic electrical test environment of the capacitor under test; Step b: Based on the basic electrical test environment, apply electrothermal stress superposition excitation to the capacitor under test to obtain the environmental response signal of the capacitor under test; Step c: Switch the constant current source of the environmental response signal to obtain the initial voltage value of the capacitor under test, and perform voltage trajectory tracking on the initial voltage value to obtain the charge and discharge characteristic curve of the capacitor under test; Step d: Identify the segmented slopes of the charging saturation region and the discharging cutoff region in the charge-discharge characteristic curve to obtain the internal dielectric polarization characteristics of the capacitor under test. Step e: Perform feature weighted fusion of the internal dielectric polarization characteristics and the ripple component in the environmental response signal to obtain the multidimensional fault diagnosis vector of the capacitor under test; Step f: Perform interactive feedback on the multidimensional fault diagnosis vector to achieve closed-loop detection of the capacitor under test.

Images