Real-time data processing method and system for electrical control equipment

Abstract

This application relates to the field of electrical control processing technology, and discloses a method and system for real-time data processing of electrical control equipment. The method includes: extracting time-series features from a preliminary smoothed signal and performing normalization processing to generate a standardized time-series feature description vector; retrieving historical records similar to the time-series feature description vector, constructing a reference interference pattern, comparing and correcting the preliminary smoothed signal with the reference interference pattern, performing segmented differential filtering, and generating a complete smoothed signal sequence; extracting unique interference features, associating and storing these features with the complete smoothed signal sequence, and updating the historical process database; verifying the time-domain and frequency-domain features of the complete smoothed signal sequence, converting it into a standardized current and voltage data sequence, and completing real-time data processing. If the verification fails, the process reverts to the preliminary denoising stage for reprocessing. This application improves the control accuracy and operational stability of electrical control equipment.

Description

Technical Field

[0001] This application relates to the field of electrical control processing technology, and in particular to a method and system for real-time data processing of electrical control equipment. Background Technology

[0002] Electrical control equipment is a core component of modern industrial automation systems. Its operational stability and control precision directly determine production efficiency and equipment safety. Accurate acquisition and processing of current and voltage signals at the moment of contact opening and closing are crucial for ensuring normal equipment operation. Currently, when electrical control equipment contacts actuate, mechanical vibrations can cause contact jitter, while electric arcing can induce drastic fluctuations in current and voltage. The interplay of these two types of interference causes transient signals to exhibit highly nonlinear characteristics, posing a significant challenge to signal processing.

[0003] Existing signal processing methods mostly employ fixed-parameter filtering, lacking the ability to effectively identify and isolate sudden noise, making it difficult to accurately capture complex interference patterns. Furthermore, their processing performance is inconsistent across different voltage levels and operating conditions, easily leading to a large amount of interference mixed in the acquired signals. Distorted signals not only fail to accurately reflect the actual operating status of the equipment but also interfere with the generation of subsequent control commands, causing equipment malfunctions, operational instability, and even affecting the reliability of the entire industrial system. Therefore, developing a real-time data processing method for electrical control equipment that can accurately adapt to the transient signal characteristics of contacts and achieve efficient interference elimination has become an urgent technical problem to be solved. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a real-time data processing method and system for electrical control equipment, used to achieve high-quality real-time processing of contact signals during the operation of electrical control equipment.

[0005] In a first aspect, this application provides a real-time data processing method for electrical control equipment, the method comprising: The current and voltage signals of electrical control equipment contacts are collected by sensors at the moment of opening and closing to obtain transient signal sequences. The transient signal sequences are subjected to preliminary denoising processing to generate preliminary smoothed signals. The temporal features of the preliminary smoothed signals are extracted and normalized to generate standardized temporal feature description vectors. Retrieve historical records similar to the time-series feature description vector from a pre-established historical process database, construct a reference interference pattern for the current data processing process based on the historical records, compare the preliminary smoothed signal with the reference interference pattern for interference comparison and error correction, and then perform segmented differential filtering for different interference types to generate a complete smoothed signal sequence. Residual analysis is performed on the complete smooth signal sequence and the transient signal sequence to extract unique interference features. The unique interference features are associated with and stored with the complete smooth signal sequence, and the historical process database is updated. The complete smooth signal sequence is verified in both the time and frequency domains. If the verification is successful, the complete smooth signal sequence is converted into a standardized current and voltage data sequence. Real-time data processing is then performed based on the standardized current and voltage data sequence. If the verification fails, the process is reverted to the initial denoising stage and reprocessed until the verification is successful.

[0006] Secondly, this application provides a real-time data processing system for electrical control equipment, the system comprising: The feature extraction unit is used to collect the current and voltage signals at the moment of opening and closing of the contacts of the electrical control equipment through the sensor to obtain the transient signal sequence, perform preliminary denoising processing on the transient signal sequence to generate a preliminary smooth signal, extract the time series features from the preliminary smooth signal and perform normalization processing to generate a standardized time series feature description vector. The pattern construction unit is used to retrieve historical records similar to the time-series feature description vector from a pre-established historical process database, construct a reference interference pattern for the current data processing process based on the historical records, perform interference comparison and error correction processing on the preliminary smoothed signal and the reference interference pattern, and then perform segmented differential filtering for different interference types to generate a complete smoothed signal sequence. The signal analysis unit is used to perform residual analysis on the complete smooth signal sequence and the transient signal sequence, extract unique interference features, associate and store the unique interference features with the complete smooth signal sequence, and update the historical process database. The real-time processing unit is used to perform time-domain and frequency-domain feature verification on the complete smooth signal sequence. After the verification is successful, the complete smooth signal sequence is converted into a standardized current-voltage data sequence. Real-time data processing is completed based on the standardized current-voltage data sequence. If the verification fails, the process is backtracked to the preliminary denoising stage for reprocessing until the verification is successful.

[0007] Compared with the prior art, the beneficial effects of the present invention are at least as follows: First, dual-channel current and voltage signals are acquired simultaneously. Contact jitter spikes and continuous arcing oscillations are specifically removed. By comparing the dual-channel denoising results and iteratively adjusting parameters, residual noise is effectively reduced, improving the signal-to-noise ratio. Simultaneously, local noise distribution characteristics are recorded to enhance the targeting and flexibility of denoising. Based on envelope detection, each stage of contact action is accurately segmented. The amplitude distribution and pulse interval patterns of interference are extracted and a time-series feature vector is constructed. Normalization eliminates dimensional differences, ensuring the comparability and accuracy of features, achieving a comprehensive and quantitative characterization of contact transient interference features, providing accurate input for historical record retrieval. Simultaneously, historical similar records are retrieved using Euclidean distance, extracting three core features: phase shift, amplitude-voltage polarity correspondence, and current zero-crossing time-series coupling degree. Through normalization, dynamic weighting, and real-time feature verification, a reference interference pattern adapted to the current operating condition is constructed. Adaptation tags such as equipment type and voltage level, along with contact action stage characteristics, are incorporated to achieve a high degree of matching between the interference pattern and the actual operating condition. By comparing signal deviations point by point, a Kalman filter is activated to recursively estimate the state of random disturbances in contact jitter, effectively suppressing irregular disturbances. A support vector machine is used to perform binary classification on continuous abnormal deviation segments, accurately distinguishing between real fault signals and residual interference, and marking reliable signal subsequences to avoid fault misjudgment and interference residue, thus improving the reliability of signal processing. Then, an adaptive weighted average filter with a forgetting factor that increases with the interference intensity is applied to the arc interference-dominant segment, and a median sorting depth filter that is positively correlated with the jitter pulse density is applied to the contact jitter-dominant segment, achieving accurate and adaptive suppression of different types of interference while balancing signal smoothness and preservation of real features. After signal splicing and boundary linear interpolation / weighted averaging for smooth transition, the signal's temporal continuity is ensured. Smoothness verification and parameter iterative adjustment further optimize signal quality. By calculating the point-by-point residuals, unique interference features such as the segmented probability density of interference duration and the local peak attenuation rate are extracted and associated with a complete smooth signal sequence for storage. The reference interference mode is then updated with weights, creating a self-learning and self-optimizing closed loop in the database. The accuracy of interference mode matching in subsequent processing continuously improves with the number of processing iterations, enhancing the method's adaptability and robustness. Finally, the signal is comprehensively validated using time-domain features such as average amplitude and peak value, as well as frequency-domain features such as dominant frequency components and noise band energy. If a validation fails, the process reverts to the initial denoising stage to adjust parameters, ensuring the quality of the output signal. Standardized data is generated through data format conversion and unified sampling rate, directly mapped to control command input signals, achieving seamless adaptation with electrical control equipment control systems. Simultaneously, the standardized data is stored in the database, enriching the historical data reserve. Attached Figure Description

[0008] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are 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 This is a flowchart of a real-time data processing method for electrical control equipment according to an embodiment of this application; Figure 2 This is a schematic diagram of temporal feature extraction according to an embodiment of this application; Figure 3 This is a schematic diagram comparing the segmented differential filtering effects of embodiments of this application; Figure 4 This is a schematic diagram of the structure of a real-time data processing system for electrical control equipment according to an embodiment of this application. Detailed Implementation

[0010] This application provides a real-time data processing method and system for electrical control equipment. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data used can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0011] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of a real-time data processing method for electrical control equipment in this application includes: Step S1: Collect the current and voltage signals at the moment of opening and closing of the contacts of the electrical control equipment through the sensor to obtain the transient signal sequence. Perform preliminary denoising processing on the transient signal sequence to generate a preliminary smooth signal. Extract the time series features from the preliminary smooth signal and perform normalization processing to generate a standardized time series feature description vector.

[0012] The process of generating a preliminary smoothed signal includes: The sensor synchronously acquires dual-channel current and voltage signals at the moment of contact opening and closing. For the spike pulses caused by contact jitter and the continuous oscillations caused by arcing in the dual-channel current and voltage signals, a wavelet threshold denoising method is used to process them to generate a preliminary smoothed transient signal sequence. If there is local noise in the transient signal sequence, the noise distribution characteristics are recorded. At the same time, the denoising results of the dual-channel current and voltage signals are compared to determine whether the denoising effect meets the preset standard. If it does not meet the standard, the denoising parameters are adjusted and reprocessed until the denoising effect meets the preset standard, and the final preliminary smoothed signal is generated.

[0013] The generation of standardized time-series feature description vectors includes: A time-series segmentation method based on envelope detection is used for the initial smoothed signal to determine the time nodes of the start of mechanical action, initial contact, stable closure, and arc extinction. The amplitude distribution characteristics of the interference waveform envelope at the boundary points of each time node are extracted, and the regularity data of the repetition interval of the interference pulse are statistically analyzed. A time-series feature description vector is constructed based on the amplitude distribution characteristics and regularity data. The data of each dimension of the time-series feature description vector are normalized to generate a standardized time-series feature description vector.

[0014] Specifically, the transient signal sequence is a set of current and voltage signal data obtained by continuous sampling by the sensor at the moment the contact closes or opens. It includes real signals generated by the mechanical action of the contact and the arcing process, while also containing interference signals such as spike pulses caused by contact jitter and continuous oscillations caused by arcing.

[0015] The dual-channel current and voltage signals are the current and voltage signals acquired by the current transformer and voltage sensor respectively at the moment of contact opening and closing. The two signals are sampled synchronously to ensure consistency between the current and voltage signals in the time dimension, fully reflecting the coordinated change characteristics of current and voltage at the moment of contact opening and closing. The wavelet threshold denoising method is a signal denoising algorithm based on wavelet transform. Its core principle is to decompose the original signal into low-frequency approximate components and high-frequency detail components of different scales through wavelet transform. The low-frequency approximate components correspond to the true characteristics of the signal, while the high-frequency detail components mainly contain noise interference. By setting a threshold to process the high-frequency detail components and removing the noise components, the signal is reconstructed through inverse wavelet transform, achieving noise elimination and signal smoothing. In this embodiment, the input data for the wavelet threshold denoising method is the original transient signal of the dual-channel current and voltage signals. First, the Daubechies wavelet is selected as the wavelet basis function for the dual-channel signal. Multi-scale decomposition is performed to separate high-frequency noise components generated by contact jitter and arcing. Then, a soft thresholding function is applied to the decomposed high-frequency coefficients. The soft thresholding function sets the components whose absolute value is less than the threshold to zero and subtracts the threshold from the components whose absolute value is greater than the threshold before retaining them. The threshold is adaptively calculated based on the noise level of the signal. The calculation method is to multiply the noise standard deviation by the square root of 2 and the square root of the logarithmic signal length. Finally, the processed wavelet coefficients are reconstructed to output a preliminary smoothed transient signal sequence. This preliminary smoothed transient signal sequence has initially eliminated spike pulses and continuous oscillation interference, while retaining the core temporal characteristics of the signal.

[0016] Local noise refers to residual noise interference in localized areas after initial denoising. It often manifests as amplitude anomalies in the signal. Noise distribution characteristics include the location, average interval, and amplitude histogram of these anomalies, reflecting the distribution pattern of residual noise. To determine whether the denoising effect meets the preset standard, the root mean square error (RMSE) of the current and voltage channel signals before and after denoising is compared. The RMSE is the square root of the average of the sum of the squares of the differences between the sampling points before and after denoising. It is a core indicator for measuring the degree of distortion and residual noise after denoising. If the RMSE is less than the preset threshold, the denoising effect is considered to meet the preset standard. If the RMSE is greater than or equal to the preset threshold, the denoising effect is considered to fail to meet the preset standard. The denoising parameters include wavelet decomposition levels and threshold coefficients. The decomposition accuracy of high-frequency noise can be improved by increasing the wavelet decomposition levels, or the threshold coefficients can be adjusted to enhance the removal of noise components. The wavelet threshold denoising process after parameter adjustment still uses the original current and voltage dual-channel signals as input, repeating the above-mentioned multi-scale decomposition, soft thresholding, and signal reconstruction process until the denoising effect meets the preset standard, generating the final preliminary smooth signal. This preliminary smooth signal is a signal sequence without obvious noise interference and retains the core timing characteristics of the contact opening and closing.

[0017] Envelope detection is a signal processing method that extracts the envelope of a signal to reflect the trend of signal amplitude changes. In this embodiment, Hilbert transform is used to implement envelope detection. The principle is to perform Hilbert transform on the initially smoothed signal to obtain an analytic signal. The magnitude of the analytic signal is the instantaneous amplitude of the signal. Connecting the continuous instantaneous amplitude values ​​forms the signal envelope, including the upper envelope and the lower envelope, which can clearly reflect the overall change law of the signal amplitude. The input data of the time sequence segmentation method based on envelope detection is the initially smoothed signal, and the output data are the start time of the contact mechanical action, the initial contact time of the contact, and the contact... The four key time nodes—the stable closing moment and the arc extinction moment—are segmented as follows: The judgment is based on the amplitude change threshold of the envelope. When the envelope amplitude first exceeds the preset starting threshold, it is marked as the start of the mechanical action of the contact. When the envelope amplitude experiences its first sharp drop followed by a rebound, it is marked as the initial contact moment. When the envelope amplitude stabilizes within a low fluctuation range, it is marked as the stable closing moment of the contact. When the envelope amplitude decays to near zero, it is marked as the arc extinction moment. Determining these four time nodes allows for the segmentation of the complete contact opening and closing process, providing time boundaries.

[0018] The boundary points of each time node are the signal sampling points corresponding to the four key time nodes mentioned above. The amplitude distribution characteristics of the interference waveform envelope are the statistical characteristics of the amplitude of the interference waveform envelope within the local area where each boundary point is located, including the mean and variance of the amplitude distribution. The mean reflects the average level of the interference amplitude in the area, and the variance reflects the dispersion of the interference amplitude in the area, which can quantify the amplitude characteristics of the interference at different stages. The interference pulse repetition interval is the time difference between the peak values ​​of adjacent interference pulses during the contact opening and closing process. The regularity data of the interference pulse repetition interval is the statistical characteristics of this time difference, including the average interval, standard deviation, and repetition rate. The repetition rate is the proportion of the interval value falling within the preset range, which can reflect the time distribution pattern of the interference pulse and quantify the temporal characteristics of the interference at different stages. After extracting the amplitude distribution features and regularity data, the time-series feature description vector is a multi-dimensional data vector formed by concatenating the amplitude distribution features and regularity data as different dimensions. Each dimension corresponds to a specific interference feature during the opening and closing of the contact, which can comprehensively and quantitatively characterize the overall interference features at the moment of contact opening and closing. The input of this time-series feature description vector is the extracted amplitude distribution features and regularity data, and the output is the original feature vector containing multi-dimensional interference features.

[0019] For example, Figure 2 This diagram illustrates the process of extracting time-series features from a pre-smoothed signal and constructing a standardized time-series feature description vector. Figure 2In the diagram, the thick black line represents the signal envelope extracted through Hilbert transform, reflecting the amplitude variation trend of the current and voltage signals during contact opening and closing. The envelope clearly presents the overall outline of the signal, eliminates local fluctuation interference, and provides a stable amplitude variation basis for timing segmentation.

[0020] The normalization process employs the minimum-maximum normalization method. This method maps each dimension of data to a fixed interval of 0-1, eliminating the imbalance of feature weights caused by differences in units and numerical ranges across different dimensions. The calculation formula is as follows: Where X represents the raw data for a certain dimension. The minimum value of the data in this dimension. The maximum value of the data in this dimension. The input data for normalization is the original time-series feature description vector, and the output data is the normalized time-series feature description vector, with each dimension mapped to the 0-1 interval. This normalized time-series feature description vector eliminates the dimensional differences between different feature dimensions, ensuring that each feature dimension has the same numerical scale, thus ensuring the consistency of the weights of each feature dimension during the retrieval process and improving the accuracy and rationality of the retrieval results.

[0021] The above process completes the initial denoising and temporal feature extraction of transient signals at the moment of opening and closing of electrical control equipment contacts. The wavelet threshold denoising method used can specifically eliminate typical interferences caused by contact jitter and arcing. The extracted standardized temporal feature description vector can comprehensively, accurately, and quantitatively characterize the interference features of the contact opening and closing process. Moreover, the entire processing process is highly compatible with the characteristics of transient signals of electrical control equipment contacts, taking into account both real-time performance and accuracy, and can meet the actual needs of real-time data processing of electrical control equipment.

[0022] Step S2: Retrieve historical records similar to the time-series feature description vector from the pre-established historical process database, construct a reference interference pattern for the current data processing based on the historical records, compare the preliminary smoothed signal with the reference interference pattern for interference and error correction, and then perform segmented differential filtering for different interference types to generate a complete smoothed signal sequence.

[0023] The reference interference pattern for the current data processing process includes: Based on the Euclidean distance between the time-series feature description vector and the feature vectors of each historical record in the historical process database, several sets of historical records with similarity higher than a preset similarity threshold are extracted. From these sets of historical records, the phase shift data relative to the contact action at the time of interference occurrence, the correlation between interference amplitude and voltage polarity, and the temporal coupling degree between interference and current zero crossing point are extracted. Based on the phase shift data, correlation, and temporal coupling degree, a reference interference mode adapted to the current data processing process is constructed.

[0024] The reference interference pattern for constructing the current data processing procedure also includes: The phase offset data, related correspondences, and temporal coupling degree are normalized to generate corresponding normalized features. Weight values ​​corresponding to the normalized features are obtained based on the feature stability coefficients in historical records, and these weight values ​​are adaptively adjusted according to equipment operating conditions. The normalized features are constructed into multi-dimensional feature vectors. The multi-dimensional feature vectors are weighted based on the adjusted weight values ​​to obtain core feature values. A basic reference interference pattern is constructed by combining the core feature values, adjusted weight values, and historical interference type labels. Real-time features of the current preliminary smoothed signal are extracted and normalized to obtain a real-time feature vector. The deviation between the real-time feature vector and the feature vector of the basic reference interference pattern is calculated. If the deviation exceeds a preset deviation threshold, the search history is expanded and a pattern is constructed. Otherwise, the basic reference interference pattern is split into feature dimension layers according to the contact action stage, and the feature value range of each stage is labeled. Adaptation tags containing equipment type, voltage level, and contact opening / closing type are added to generate a reference interference pattern adapted to the current data processing process.

[0025] The process of comparing the initial smoothed signal with a reference interference mode and performing error correction includes: The preliminary smoothed signal is compared point by point with the reference interference pattern. If the amplitude distribution characteristics of the interference waveform envelope of a local segment deviate from the preset range, the Kalman filter is activated to recursively estimate the state of the random disturbance dominated by contact jitter in the local segment, generating a preliminary estimated smoothed trajectory signal. The segmented deviation sequence is calculated between the smoothed trajectory signal and the reference interference pattern. If there are abnormal deviations in the segmented deviation sequence that exceed the preset length, the signal features of the deviation segment are extracted and binary classification is performed using a support vector machine. The reliable signal subsequence is marked according to the classification results.

[0026] Generating a complete smooth signal sequence includes: Align the timing of the trusted signal subsequence with the timing of the interference stage in the reference interference mode to obtain the interference type marked in the reference interference mode. If the interference type is dominated by arc interference, an adaptive weighted average filtering process with the forgetting factor increasing with the interference intensity is applied to the corresponding signal stage. If the interference type is dominated by contact jitter, a median sorting depth filtering process positively correlated with the jitter pulse density is applied to the corresponding signal stage. The filtered signals of each stage are spliced ​​together according to the original timing and the splicing boundary is smoothed to generate a complete smooth signal sequence. At the same time, the smoothness of the complete smooth signal sequence is verified. If the verification fails, the filtering parameters are adjusted and the filtering is repeated until the verification passes.

[0027] Specifically, this step revolves around retrieving similar records from the historical process database, constructing reference interference patterns, comparing and correcting signal interference, and segmented differential filtering. Through multi-dimensional feature analysis and adaptive signal processing algorithms, it achieves precise interference elimination of transient signals from the opening and closing of electrical control equipment contacts, generating a complete and smooth signal sequence that fits the actual operating conditions of the equipment. The entire processing is highly adapted to the dynamic interference characteristics of the contact signals, taking into account both the accuracy and real-time performance of signal processing.

[0028] The historical process database stores records of various interference signals processed during the opening and closing of contacts in electrical control equipment. Each record contains a temporal feature description vector, interference feature data, and interference type labeling for the corresponding contact action process. It serves as the fundamental data carrier for matching similar interference patterns. The temporal feature description vector is a multi-dimensional feature vector extracted from the initial smoothed signal of the contact opening and closing, after normalization. Its dimensions represent the interference waveform and pulse distribution characteristics at different stages of the contact action, serving as the core basis for retrieving similar historical records. Euclidean distance is a numerical indicator representing the similarity between two multi-dimensional feature vectors. It is calculated by summing the squares of the differences in each dimension of the two feature vectors and taking the square root. A smaller Euclidean distance indicates a higher similarity between the two feature vectors. The preset similarity threshold is a similarity judgment value set according to the actual signal processing requirements of the equipment. This threshold allows filtering out historical records highly similar to the interference characteristics of the current contact action. Phase offset data is the time difference between the occurrence time of the interference signal and the start time of the contact mechanical action in historical records. It can characterize the occurrence pattern of the interference signal in the contact action sequence. The correlation between interference amplitude and voltage polarity is the relationship between the magnitude of the interference signal amplitude in historical records and the positive or negative polarity of the voltage at the corresponding time. It is quantified by calculating the ratio of the average interference amplitude in the positive polarity period to the average interference amplitude in the negative polarity period. Voltage polarity refers to the positive or negative polarity of the voltage, which is divided into positive and negative polarity. This feature can reflect the influence of voltage polarity on the interference amplitude. The temporal coupling degree between interference and current zero-crossing point is the time correlation characteristic between the occurrence time of interference signal and current zero-crossing point in historical records. It is measured by calculating the average value of the absolute value of the time difference between the peak time of interference and the most recent current zero-crossing point. The current zero-crossing point refers to the zero value point that the current signal passes through when the value changes from positive to negative or from negative to positive. This feature can characterize the coupling law between interference signal and current periodic change. The above three types of data are the core feature data for constructing reference interference mode, which can reflect the interference characteristics in the contact opening and closing process from different dimensions.

[0029] First, the phase offset data, the correlation correspondence, and the temporal coupling degree are normalized respectively. Normalization is a signal processing method to eliminate the imbalance of feature weights caused by different dimensions and numerical ranges of different feature data. In this step, an appropriate normalization method is adopted for the value characteristics of different feature data, and all types of feature data are mapped to a fixed interval of 0-1, so that feature data of different dimensions have a unified numerical scale. The weight values ​​corresponding to normalized features are obtained based on the feature stability coefficients in historical records. These weight values ​​are then adaptively adjusted in conjunction with equipment operating conditions. The feature stability coefficient is an index calculated based on the variance of the normalized feature in several extracted historical records. The smaller the variance, the more stable the feature is in similar interference processes and the higher its representational value for interference patterns. The reciprocal of the feature stability coefficient is the initial weight value of the feature. All initial weight values ​​are then normalized to obtain the basic weight values ​​for each normalized feature. Equipment operating conditions refer to the current operating status of electrical control equipment, including the equipment's voltage level, contact opening and closing type, and load conditions. Contact opening and closing type refers to the action type of the electrical control equipment contacts, which is divided into contact closing type and contact opening type. This is the most core action condition feature of the contacts. The basic weight values ​​are adaptively adjusted in conjunction with the equipment operating conditions to make the weight values ​​more consistent with the current signal processing requirements of the equipment, thereby strengthening the influence of high-value features in the construction of reference interference patterns.

[0030] Specifically, the multidimensional feature vector is a vector formed by concatenating normalized features as different dimensions. It can completely represent the three types of core interference features. The core feature value is the value obtained by weighting and summing the data of each dimension of the multidimensional feature vector according to the corrected weight values. It can comprehensively reflect the overall matching degree between the current contact action interference features and historical similar interference features. The historical interference type label is the classification identifier of interference type in the historical record, mainly including two types: arc interference-dominated and contact jitter-dominated. Arc interference-dominated means that the signal interference during the contact opening and closing process is mainly caused by the continuous oscillation generated by the arc. Contact jitter-dominated means that the signal interference during the contact opening and closing process is mainly caused by the spike pulse generated by the mechanical vibration of the contact. The basic reference interference mode is the preliminary interference mode that integrates historical similar interference features.

[0031] The real-time features of the current preliminary smoothing signal are the phase shift trend, amplitude-polarity correlation, and temporal coupling degree extracted from the current preliminary smoothing signal, which correspond to the features of the constructed reference interference pattern. After the same normalization process as described above, a real-time feature vector is formed, which can characterize the actual interference features of the current contact action process. The deviation value is calculated using Euclidean distance and is used to characterize the matching degree between the basic reference interference pattern and the current actual interference features. If the deviation value exceeds the preset deviation threshold, it indicates that the basic reference interference pattern is not well adapted to the current actual working condition. At this time, the search range of historical records is expanded, the number of historical records searched is increased, and the above feature extraction, normalization, weighted calculation, and pattern construction process is repeated until the deviation value between the constructed basic reference interference pattern and the current real-time features is lower than the preset deviation threshold. If the deviation value does not exceed the preset deviation threshold, it indicates that the basic reference interference pattern is well adapted to the current actual working condition. At this time, the basic reference interference pattern is processed according to the contact action order. The feature dimension layer is split into segments and the feature value range of each stage is marked. The contact action stage includes the contact mechanical action initiation stage, the initial contact stage, the stable closing stage, and the arc extinguishing stage. Splitting the feature dimension layer according to this stage enables the reference interference mode to accurately match the interference characteristics of different stages of contact action. Then, an adaptation label containing equipment type, voltage level, and contact opening and closing type is added to the split interference mode. This adaptation label can clearly identify the applicable working conditions of the reference interference mode. Finally, a reference interference mode adapted to the current data processing process is generated. This reference interference mode is a multi-dimensional interference feature mode that integrates historical similar data with current real-time features and fits the actual working conditions of the equipment.

[0032] Specifically, the amplitude and phase of each sampling point of the preliminary smoothed signal are compared one by one with the characteristic values ​​of the corresponding time position in the reference interference mode, and the point-by-point deviation value is calculated. Through this point-by-point comparison, the signal segments in the preliminary smoothed signal that deviate from the reference interference mode can be accurately identified. The interference waveform envelope is the amplitude change contour of the interference signal extracted by the envelope detection method. Its amplitude distribution characteristics are the statistical characteristics such as the mean and variance of the contour. The preset range is the deviation judgment threshold set according to the signal processing accuracy requirements of the equipment. The Kalman filter is a recursive estimation algorithm based on minimum mean square error. Its core principle is to recursively estimate the state of a dynamic system through two steps: prediction and update. First, it predicts the system state at the current moment based on the state estimate from the previous moment. Then, it corrects the prediction by incorporating the measured value at the current moment, obtaining the optimal state estimate for the current moment. The algorithm's input data consists of a local signal segment with random disturbances caused by contact jitter, and preset process noise covariance matrices and measurement noise covariance matrices. The process noise covariance matrix characterizes the uncertainty of system state changes and can be set as a diagonal matrix, with diagonal elements determined based on the maximum possible acceleration of contact jitter. The measurement noise covariance matrix characterizes the error level of the measurement data and is set according to the sensor's nominal accuracy. The Kalman filter output data is the signal data after recursive state estimation. Based on this output data, a preliminary estimated smooth trajectory signal is generated. This smooth trajectory signal has initially eliminated the random disturbances caused by contact jitter, making the signal trajectory more closely resemble the actual contact action characteristics.

[0033] The smooth trajectory signal is segmented according to a preset number of sampling points. The Euclidean distance between each signal segment and the corresponding time segment in the reference interference mode is calculated. The distance values ​​of all segments are arranged in time sequence to form a segmented deviation sequence, which can characterize the matching degree between each segment of the smooth trajectory signal and the reference interference mode. If the distance values ​​of multiple consecutive segments in the segmented deviation sequence exceed a preset deviation threshold, the signal features of the deviating segments are extracted and binary classification is performed using a support vector machine. The signal features of the deviating segments include multidimensional data that can characterize the interference features of the deviating segments, such as the mean amplitude distribution, the variance of the pulse interval, and the phase shift. The support vector machine is a supervised learning classification algorithm based on maximizing the classification boundary. Its core principle is to map the feature data in the low-dimensional space to the high-dimensional space through the kernel function, and construct the optimal classification hyperplane in the high-dimensional space to achieve accurate classification of different categories of data. In this step, the radial basis function is used as the kernel function. The support vector machine (SVM) can effectively achieve high-dimensional mapping of low-dimensional features, improving classification accuracy. The input data of the SVM consists of the extracted deviation segment signal features and the classification model trained on historical data. The classification model is trained using real fault signals and residual interference signals labeled in the historical database. The output data of the SVM consists of category labels, which are divided into real fault signals and residual interference signals. Based on the classification results, a reliable signal subsequence is marked. That is, only deviation segments classified as residual interference signals and signal segments with a high degree of matching with the reference interference pattern are marked as reliable signal subsequences. This reliable signal subsequence serves as the effective signal basis for subsequent differential filtering.

[0034] Specifically, each sampling point of the reliable signal subsequence is matched with the temporal position of each interference stage in the reference interference mode according to the time dimension, and then the interference type marked in the reference interference mode is obtained, so that each signal segment corresponds to a clear interference type identifier. If the interference type is dominated by arc interference, the forgetting factor is a parameter that characterizes the weight of historical output values ​​during the filtering process, and its value ranges from 0 to 1. The interference intensity is the standard deviation of the amplitude of all sampling points within a sliding window centered on the signal sampling point. This standard deviation can effectively characterize the intensity of arc interference oscillation. The forgetting factor increases with the interference intensity, that is, the greater the interference intensity, the closer the forgetting factor value is to 1. The core principle of adaptive weighted average filtering is to achieve signal smoothing through recursive calculation. The filtered output value of the k-th sampling point in the signal sequence is the product of the forgetting factor and the filtered output value at the previous moment, plus the product of the difference between 1 and the forgetting factor and the current sampling value. The input data of the adaptive weighted average filtering method is the reliable signal subsequence of the arc interference-dominated stage and the forgetting factor calculated according to the interference intensity. The output data is the arc interference stage signal after smoothing. Through this filtering process, the smoothing effect can be enhanced in the region of high interference intensity, effectively suppressing the continuous oscillation caused by arcing, improving the tracking performance of the signal in the region of low interference intensity, and preserving the true signal characteristics. If the interference type is contact jitter-dominated, the jitter pulse density is the ratio of the number of jitter pulse events to the length of the statistical time window centered on the signal sampling point. Jitter pulse events are identified by calculating the first-order difference of the signal; points where the absolute value of the difference exceeds a preset threshold are identified as pulse edge points. The region between two consecutive edge points constitutes a jitter pulse event. The core principle of median sorting depth filtering is to sort all sampled values ​​within a preset window size before and after the signal sampling point, and take the median of the sorted values ​​as the filtered output value for that sampling point. The window size is positively correlated with the jitter pulse density; that is, the higher the jitter pulse density, the larger the filtering window size. The input data for the median sorting depth filtering method are a reliable signal subsequence of the contact jitter-dominated stage and an odd-numbered window size determined based on the jitter pulse density. The odd-numbered window size ensures the unique existence of the median. The output data of this filtering method is the smoothed contact jitter stage signal. This filtering process enhances the suppression of spike pulses in areas with dense jitter pulses and reduces excessive smoothing of the signal in areas with sparse pulses, preserving the true temporal characteristics of the contact action.

[0035] Following the temporal sequence of contact actions, the filtered signals from the arc interference-dominated stage, the contact jitter-dominated stage, and the stage without significant interference are spliced ​​together. Then, at the splicing boundary where amplitude jumps occur, linear interpolation or weighted averaging is used to process a small number of sampling points before and after the boundary, ensuring the spliced ​​signal has a continuous amplitude without jumps. Linear interpolation uses the splicing boundary as the dividing point, taking the last valid sampling point before the boundary and the first valid sampling point after the boundary as interpolation anchor points. Based on the amplitude of these two anchor points and the sampling interval, a linear function is used to construct the amplitude change relationship between the two points, and the amplitude of each sampling point within the jump interval is calculated point by point. The first method ensures a linear and continuous change in amplitude between anchor points, eliminating the step-like characteristics of abrupt amplitude changes. The second method involves selecting a preset number of sampling points before and after the splicing boundary to form a processing window. The sampling points within the window are assigned distance-related weights, meaning that the closer the sampling point is to the splicing boundary, the larger the weight value, and the farther the sampling point is, the smaller the weight value. The amplitude of each sampling point within the window is recalculated by weighted summation and averaging, allowing the amplitudes on both sides of the boundary to transition smoothly through weighted transitions. Both methods reassign the amplitude values ​​of the sampling points at the transition positions to achieve continuous amplitude without abrupt changes at the splicing boundary, ensuring the integrity and smoothness of the signal in the time domain. Smoothness verification is achieved by calculating the rate of change of local curvature of the signal. The rate of change of local curvature is the average of the absolute values ​​of the second-order difference within the window calculated by the sliding window after approximating the second-order derivative of the complete smooth signal sequence. This index can characterize the severity of the bending change of the signal waveform. If the rate of change of local curvature continuously exceeds the preset threshold, it indicates that there is an abnormal fluctuation segment in the signal, and the smoothness verification fails. At this time, the filtering parameters are adjusted according to the interference type corresponding to the abnormal fluctuation segment. If it is the arc interference stage, the forgetting factor of the adaptive weighted average filter is increased. If it is the contact jitter stage, the window size coefficient of the median sorting depth filter is increased. Then, the corresponding signal stage is filtered again using the adjusted filtering parameters, and the signal splicing, boundary smoothing and smoothness verification are performed again until the smoothness verification passes, and finally a complete smooth signal sequence that meets the smoothness requirements is generated.

[0036] For example, Figure 3 The accompanying comparison chart shows the effects of segmented differentiated filtering, illustrating different filtering strategies employed for different types of interference and their respective processing results. Figure 3The upper subplot shows a comparison of the signal segment dominated by contact jitter before and after median sorting depth filtering. Contact jitter is caused by mechanical vibration of the contact, and its typical characteristics are high-amplitude spike pulses with short durations. The frequency of these pulses is related to the mechanical structure characteristics of the contact. The lower subplot shows a comparison of the signal segment dominated by arc interference before and after adaptive weighted average filtering. Arc interference is generated by gas breakdown during the opening and closing of the contact, and its typical characteristics are high-frequency continuous oscillations. The oscillation frequency changes with the development of the arc, and the amplitude shows an exponential decay trend.

[0037] The above process completes the entire workflow from historical similar record retrieval and reference interference pattern construction to signal comparison and error correction, and segmented differential filtering. The constructed reference interference pattern integrates historical data with current real-time features, exhibiting high adaptability to operating conditions. The algorithms employed, such as Kalman filter, support vector machine, adaptive weighted average filtering, and median sorting deep filtering, are all designed for specific interference characteristics in the contact opening and closing process, achieving precise elimination of two core types of interference: arcing and contact jitter. The generated complete and smooth signal sequence effectively eliminates interference signals while preserving the true temporal characteristics of contact actions. The algorithm design of the entire processing process is highly compatible with the real-time data processing requirements of electrical control equipment, ensuring the accuracy, real-time performance, and robustness of signal processing.

[0038] Step S3: Perform residual analysis on the complete smooth signal sequence and the transient signal sequence, extract unique interference features, associate and store the unique interference features with the complete smooth signal sequence, and update the historical process database.

[0039] The residual analysis of complete smooth signal sequences and transient signal sequences includes: The residual sequences are obtained by calculating the pointwise residuals of the complete smooth signal sequence and the transient signal sequence, respectively. The residual sequences are then segmented by thresholding and the interference segments are identified. The duration characteristics of each interference segment are extracted and the segmented probability density of the interference duration is statistically obtained. Local peak points within each interference segment are identified and the decay rate characteristics of the local peaks of the interference are statistically obtained. The segmented probability density and decay rate characteristics are used as unique interference features and associated with the complete smooth signal sequence to be stored in the historical process database. The reference interference modes of the corresponding categories in the historical process database are then updated with weights based on the newly stored data, thus completing the update of the historical process database.

[0040] Specifically, this step revolves around the residual analysis of complete smooth signal sequences and transient signal sequences. Through precise residual calculation and interference feature extraction, it captures the unique interference patterns of this data processing process. At the same time, it completes the associated storage and weighted update of the historical process database, allowing the reference interference patterns in the database to be continuously optimized with the actual processing process. This further improves the accuracy of interference pattern matching in data processing, enabling the entire real-time data processing method for electrical control equipment to form a self-learning and self-optimizing closed loop.

[0041] The complete smooth signal sequence is a current and voltage signal sequence generated after interference comparison, error correction processing, and segmented differential filtering, which is free of obvious interference and retains the true timing characteristics of contact opening and closing. The transient signal sequence is the original current and voltage signal sequence of the electrical control equipment at the moment of contact opening and closing collected by the sensor. It is the original data foundation of the entire data processing process. Residual analysis is a signal processing method that mines the interference characteristics in the original signal by calculating the difference between the two signal sequences. Its core function is to extract the unique interference patterns in the current contact opening and closing process, provide effective data for updating the historical process database, and verify the interference elimination effect of the smoothing process.

[0042] Pointwise residuals refer to the values ​​obtained by calculating the difference between the amplitude of each sampling point in a complete smooth signal sequence and the amplitude of the corresponding sampling point in a transient signal sequence. A sampling point refers to a discrete data point collected from a continuous current and voltage signal at a fixed time interval. The corresponding time position refers to the sampling point collected at the same moment in the two signal sequences. The calculation of pointwise residuals follows the principle of temporal consistency to ensure that each residual value can accurately reflect the amplitude difference between the original signal and the smooth signal at the corresponding moment. By arranging the pointwise residual values ​​of all moments in the temporal order of the original signal, the residual sequence can be obtained. This residual sequence is a one-dimensional numerical sequence, and the magnitude of each value represents the amplitude intensity of the interference signal at the corresponding moment. The larger the absolute value of the value, the stronger the interference at that moment. The residual sequence completely preserves the temporal and amplitude characteristics of the interference signal during the opening and closing of the contact.

[0043] Threshold segmentation is a signal processing method that classifies the values ​​in a residual sequence according to the relationship between their absolute values ​​and the threshold by setting a residual amplitude threshold. In this embodiment, the residual amplitude threshold is a fixed value set according to the signal processing accuracy requirements of the electrical control equipment, the rated operating parameters of the equipment, and historical interference processing data. Its function is to distinguish the interference signal from the normal signal fluctuation in the residual sequence. The interference segment refers to the signal segment in the residual sequence whose absolute residual value continuously exceeds the preset residual amplitude threshold. This segment corresponds to the continuous period of interference in the original transient signal sequence. When performing threshold segmentation on the residual sequence, the sequence is traversed from the beginning position. When the absolute residual value is detected to exceed the threshold for the first time, it is marked as the starting point of the interference segment. When the absolute residual value is detected to fall back below the threshold, it is marked as the ending point of the interference segment. After traversing the entire residual sequence according to this rule, all interference segments can be identified. Each identified interference segment includes information such as the start time, end time, duration, and all residual values ​​in the segment, providing a clear analysis interval for the subsequent extraction of interference features.

[0044] The duration characteristic of the interference segment refers to the length of time from the start time to the end time of each identified interference segment. This characteristic can characterize the duration pattern of different types of interference during the opening and closing of the contacts. For example, interference caused by contact jitter is usually a short-term interference segment, while interference caused by arcing is a long-term interference segment. The segmented probability density refers to dividing the duration of all interference segments into segments according to a preset time interval and calculating the proportion of the number of interference segments appearing in each time interval to the total number of interference segments. Its core function is to quantify the distribution pattern of interference duration during this data processing, which is one of the unique interference characteristics of this process. In the specific statistical process, firstly, based on the common duration range of interference during the opening and closing of electrical control equipment contacts, several continuous and non-overlapping time intervals are divided. Then, each identified interference segment is assigned to the corresponding time interval according to its duration. The number of interference segments in each time interval is counted. Finally, the number of interference segments in each time interval is divided by the total number of interference segments to obtain the probability value corresponding to each time interval. All time intervals and their corresponding probability values ​​together constitute the segmented probability density of interference duration. This segmented probability density presents the interference duration pattern of this process in the form of a statistical distribution, which can accurately reflect the difference in interference duration between this process and historical processes.

[0045] A local peak point refers to a point in the residual sequence of a single interference segment where the amplitude is greater than the amplitude of its immediate and adjacent sampling points. In other words, it's the sampling point where the interference amplitude reaches its local maximum within the interference segment. This point corresponds to the peak time of the interference intensity within the segment. Local peak point identification is achieved by traversing the residual sequence within each interference segment, comparing the amplitude of each sampling point with its immediate and adjacent sampling points. The sampling point that satisfies the maximum amplitude condition is the local peak point. If multiple local peak points exist within an interference segment, they are marked sequentially according to their temporal order. The decay rate characteristic of interference local peaks refers to the rate attenuation of the interference amplitude over time by calculating the difference between the amplitude and time for adjacent local peak points within the same interference segment. This characteristic characterizes the decay pattern of the interference intensity during this process. Different types of interference have different decay rate characteristics. For example, interference caused by electric arcing has a slower peak decay rate, while interference caused by contact jitter has a faster peak decay rate, which is another core unique interference characteristic of this process. In the specific statistical process, for interference segments with multiple local peak points, the amplitude difference between the next local peak point and the previous local peak point is calculated. Then, this amplitude difference is divided by the time interval between the two local peak points to obtain a single attenuation rate value. The attenuation rate values ​​of all adjacent local peak points within the same interference segment are statistically analyzed, and their average value, variance, and other statistical quantities are calculated. For interference segments with only one local peak point, the amplitude and time information of that peak point are directly recorded. Finally, the attenuation rate statistics of all interference segments are integrated to form the interference local peak attenuation rate characteristics of this process. This characteristic presents the unique interference pattern of this process from the perspective of the change in interference intensity.

[0046] Each record in the historical process database contains information such as the signal sequence, interference characteristics, and interference type label for the corresponding processing process. Associative storage refers to establishing a unique correspondence between the unique interference characteristics of the current process and the complete smooth signal sequence, saving them to the historical process database in key-value pair format. The key is a unique identifier for the current data processing process, containing information such as equipment number, contact opening / closing time, and equipment operating condition. The value contains data such as the complete smooth signal sequence, the segmented probability density of the interference duration, and the attenuation rate characteristics of the local interference peaks. This allows for quick retrieval of all relevant data for the current process through the unique identifier, while ensuring the correlation and integrity of the data in the database. During storage, all data is converted according to a preset standardized format to ensure consistency with the data format in the historical database, improving database management and retrieval efficiency.

[0047] The corresponding category of reference interference pattern refers to the reference interference pattern in the historical process database that is consistent with the characteristics of the current data processing process, such as the contact opening and closing type, equipment condition, and interference type. Weighted update refers to the weighted average calculation of the newly stored unique interference features and the original features in the historical reference interference pattern, and the feature data of the original reference interference pattern is updated with the new calculation result, so that the reference interference pattern can integrate the interference pattern of the current process and achieve continuous optimization. In the specific weighted update process, firstly, a reference interference pattern matching the characteristics of the current process is retrieved from the historical process database to determine the original interference feature data under this pattern. Then, weight values ​​are assigned to the newly stored interference feature data and the original interference feature data respectively. The weight value of the new data is set according to the reliability of the data and the matching degree of the equipment operating conditions, while the weight value of the original data is set according to the number of times it is stored in the database and the number of successful matches. The sum of the weight values ​​of the new data and the original data is 1. Subsequently, the weighted average of each statistic in the segmented probability density and attenuation rate features is calculated to obtain the updated interference feature data. Finally, the updated interference feature data replaces the corresponding data in the original reference interference pattern to complete the weighted update of the reference interference pattern. If there is no reference interference pattern matching the characteristics of the current process in the historical process database, a corresponding reference interference pattern is created based on the unique interference characteristics of the current process and the complete smooth signal sequence and stored in the database. By using the weighted update method described above, the reference interference patterns in the historical process database can continuously incorporate new interference patterns. The matching accuracy and adaptability of the patterns continuously improve with the number of data processing iterations, providing more accurate interference pattern references for subsequent real-time data processing of electrical control equipment. This allows the entire data processing method to form a self-learning and self-optimizing closed-loop system, further improving the accuracy and efficiency of interference elimination.

[0048] The above process completed the residual analysis of the complete smooth signal sequence and the transient signal sequence, accurately extracted the unique interference features of this data processing process, and realized the associated storage and weighted update of the historical process database. The extracted unique interference features fully presented the interference pattern of this process from two dimensions: interference duration and interference intensity attenuation. This provided effective and accurate new data for database updates. The weighted update of the historical process database allows the reference interference pattern to be continuously optimized with the actual processing process, effectively solving the problem that traditional data processing methods lack a historical data iteration mechanism and cannot continuously improve the processing effect. This enables the entire real-time data processing method of electrical control equipment to form a self-learning and self-optimizing closed loop. Subsequent signal processing can rely on the updated database to achieve more accurate interference pattern matching, further improving the accuracy and efficiency of interference elimination, providing a more reliable signal basis for the generation of control commands for electrical control equipment, and ensuring the stability and control accuracy of equipment operation.

[0049] Step S4: Perform time-domain and frequency-domain feature verification on the complete smooth signal sequence. If the verification is successful, convert it into a standardized current and voltage data sequence suitable for generating control commands and output it to complete real-time data processing. If the verification fails, backtrack to the preliminary denoising stage to adjust the parameters and reprocess until the verification is successful.

[0050] The verification of the time-domain and frequency-domain features of the complete smooth signal sequence includes: The average amplitude, peak value, duration, and time-domain features of the complete smooth signal sequence are calculated. A Fourier transform is performed on the complete smooth signal sequence to obtain the frequency domain spectrum. Based on the frequency domain spectrum, the dominant frequency component and noise band energy are extracted as frequency domain features. The time-domain features and frequency domain features are compared with corresponding preset standards. If both meet the corresponding preset standards, the complete smooth signal sequence is converted into a data format and the sampling rate is unified to generate a standardized current and voltage data sequence. At the same time, the standardized current and voltage data sequence is stored in the historical process database, and key parameters are extracted from the standardized current and voltage data sequence and mapped to the control command input signals of electrical control equipment.

[0051] Specifically, this step revolves around verifying the time-domain and frequency-domain characteristics of a complete and smooth signal sequence. By accurately verifying the multi-dimensional characteristics of the signal, the final control of signal processing quality is achieved. At the same time, the standardized conversion of the signal and the generation of control command input signals are completed. If the verification fails, the signal processing is iteratively optimized by adjusting the parameters backtracking. This ensures that the output current and voltage data sequences are fully adapted to the control command generation requirements of the electrical control equipment, providing high-quality and standardized real-time data support for the precise control of the equipment. The entire process forms a closed-loop quality control of signal processing, further improving the reliability and accuracy of real-time data processing of electrical control equipment.

[0052] A complete smooth signal sequence is a current and voltage signal sequence that eliminates interference such as contact jitter and arcing while retaining the true timing characteristics of contact opening and closing. It is the final signal result of the preceding data processing. Time-domain and frequency-domain feature verification is a signal quality detection method that extracts features from the time and frequency dimensions of the signal and compares them with preset standards. Its core function is to determine whether the complete smooth signal sequence meets the signal quality requirements of the control commands generated by electrical control equipment, ensuring the effectiveness and accuracy of the output signal from multiple dimensions. The input data of this verification method is the complete smooth signal sequence, and the output data is the feature verification result and the corresponding signal processing execution command. If the verification passes, it enters the signal standardization conversion stage; if the verification fails, it triggers the retrospective adjustment of the preceding parameters.

[0053] Temporal characteristics refer to the characteristic attributes of a signal in the time dimension. They can intuitively reflect the amplitude changes and duration patterns of the signal on the time axis and are core indicators for measuring signal stability. Average amplitude is the arithmetic mean of the amplitude values ​​of all sampling points in a complete smooth signal sequence. A sampling point refers to discrete data points collected from continuous current and voltage signals at fixed time intervals. The amplitude is the magnitude of the current or voltage corresponding to the sampling point. Average amplitude characterizes the average intensity level of the signal throughout its duration, reflecting the overall amplitude stability of the signal. Peak value is the maximum amplitude value of all sampling points in a complete smooth signal sequence, characterizing the maximum amplitude fluctuation of the signal in the time dimension and reflecting the instantaneous amplitude change limit of the signal. Duration is the length of time from the first valid sampling point to the last valid sampling point in a complete smooth signal sequence. It can be calculated by multiplying the number of sampling points by the sampling interval and characterizes the effective duration of the signal during the opening and closing of contacts, reflecting the timing matching degree between the signal and the actual action of the contacts. By calculating the average amplitude, peak value, and duration, the temporal characteristics of a complete smooth signal sequence are extracted. Each temporal characteristic presents the quality attributes of the signal in the time dimension from different perspectives.

[0054] Frequency domain characteristics refer to the characteristic attributes of a signal in the frequency dimension, reflecting the frequency distribution pattern of the signal and serving as a core indicator for measuring the degree of residual noise interference in a signal. The Fourier transform is a mathematical transformation method that converts a continuous signal in the time domain into a frequency signal in the frequency domain. Its core working principle is to decompose a complex continuous signal in the time domain into a superposition of multiple sine waves with different frequencies and amplitudes. By calculating the frequency and amplitude of each sine wave, the frequency distribution pattern of the signal is obtained. The input data for this transformation method is a complete and smooth time-domain current and voltage signal sequence, and the output data is the frequency domain spectrum of the signal. The frequency domain spectrum is a spectral distribution diagram formed by plotting frequency on the horizontal axis and the signal amplitude at the corresponding frequency on the vertical axis, which can intuitively present the proportion and intensity of each frequency component in the signal. The dominant frequency component is the frequency value corresponding to the frequency point with the largest amplitude in the frequency domain spectrum. It can characterize the main frequency characteristics of a complete and smooth signal sequence and reflect the matching degree between the signal and the rated operating frequency of the electrical control equipment. The dominant frequency component of a normal contact opening and closing signal should be consistent with the rated operating frequency of the equipment. The noise frequency band energy is the energy integral value of all frequency components within a preset high-frequency band in the frequency domain spectrum. The energy is the square value of the amplitude of the corresponding frequency point. The preset high-frequency band is a high-frequency interval set according to the operating characteristics of the electrical control equipment, in which there is no effective operating frequency. The frequency components within this interval are all frequencies corresponding to noise interference. The noise frequency band energy can characterize the total intensity of residual high-frequency noise interference in the complete and smooth signal sequence and reflect the effect of the previous denoising process. Through Fourier transform and frequency domain feature extraction, the quality analysis of the frequency dimension of the complete and smooth signal sequence is completed.

[0055] Preset standards are characteristic judgment thresholds established based on the rated operating parameters of electrical control equipment, the signal quality requirements generated by control commands, and historical signal processing data. These are divided into time-domain and frequency-domain characteristic preset standards. For example, time-domain characteristic preset standards include: the average amplitude must be within 0.5 to 1.5 times the rated operating amplitude of the equipment; the peak value must not exceed twice the average amplitude; and the duration must match the typical action duration of contact opening and closing, meaning the deviation between the duration and the typical action duration should not exceed ±20% of the typical duration. The typical duration usually refers to the entire process from the start of contact action to the complete end of the action (i.e., arc extinguishing and voltage and current stabilizing). Frequency-domain characteristic preset standards include: the main frequency component being within a reasonable range near the expected operating frequency of the equipment; and the noise band energy being less than 10% of the total signal energy. The total signal energy is the energy integral value of all frequency components in the frequency spectrum. The feature comparison process follows the one-to-one correspondence principle. The extracted average amplitude, peak value, and duration are compared with the corresponding thresholds of the preset standards for time-domain features, and the extracted main frequency components and noise band energy are compared with the corresponding thresholds of the preset standards for frequency-domain features. Only when all time-domain features and frequency-domain features meet the corresponding preset standards is the feature verification considered successful. If any feature fails to meet the corresponding preset standards, the feature verification is considered unsuccessful.

[0056] Backtracking adjustment refers to returning to the initial denoising stage of step S1 when the signal quality does not meet the requirements. It involves re-executing all data processing steps through iterative optimization by adjusting the denoising parameters. Its core function is to solve the quality problems in the signal by adjusting the parameters, so as to ensure that the final output signal meets the quality requirements. First, based on the type of feature that failed the test, the root cause of the quality problem in the signal is located. If the peak value in the time domain feature is too high, it indicates that there is still residual spike pulse interference caused by contact jitter in the signal. If the noise band energy in the frequency domain feature exceeds the standard, it indicates that there is still residual high-frequency continuous oscillation interference caused by arcing in the signal. Then, the wavelet threshold denoising parameters in the initial denoising process are adjusted according to the root cause of the problem. The wavelet threshold denoising parameters include wavelet decomposition level and threshold coefficient. For the problem of excessively high peak value, the threshold coefficient can be increased to enhance the removal of spike pulse. For the problem of excessive noise band energy, the wavelet decomposition level can be increased to improve the decomposition and filtering accuracy of high-frequency oscillation. At the same time, the parameters can be adjusted differently according to the difference in signal type. For noise dominated by voltage signals, the lower limit of the threshold is adjusted first, and for noise dominated by current signals, the wavelet decomposition level is increased. After the parameters are adjusted, the initial denoising process is re-executed with the adjusted wavelet threshold denoising parameters. Then, all data processing steps such as time-series feature extraction, reference interference mode construction, interference comparison and error correction, segmented differential filtering, and residual analysis are completed in sequence to generate a new complete smooth signal sequence. The time-domain and frequency-domain feature verification of this step is then performed again. This iterative adjustment process can be executed multiple times until the time-domain and frequency-domain features of the generated complete smooth signal sequence meet the corresponding preset standards, ensuring that the final output signal has qualified quality attributes.

[0057] Standardization processing refers to the signal processing method that converts a complete, smooth signal sequence that meets quality requirements into a data format and sampling rate compatible with the control command generation system of electrical control equipment. Its core function is to achieve compatibility between the signal and the control command generation system, ensuring that the data can be effectively recognized and retrieved by the system. Data format conversion transforms the raw data format of the complete, smooth signal sequence into a floating-point array format supported by the control command generation system. Floating-point arrays accurately represent the amplitude values ​​of current and voltage while possessing high computational and retrieval efficiency, making them the standard format for data transmission and processing in industrial control equipment. Sampling rate unification adjusts the sampling rate of the complete, smooth signal sequence to the rated sampling rate of the control command generation system. The sampling rate refers to the number of signal sampling points per unit time. A unified sampling rate ensures that the temporal resolution of the signal matches the system's processing capability, avoiding timing deviations caused by sampling rate mismatches and guaranteeing the timing accuracy of control command generation. Through this dual processing of data format conversion and sampling rate unification, the complete, smooth signal sequence is converted into a standardized current and voltage data sequence. This sequence possesses acceptable signal quality and is fully compatible with the control command generation system, serving as the final output data for this real-time data processing of the electrical control equipment. The generated standardized current and voltage data sequences are associated with information such as equipment operating conditions, interference characteristics, and adjusted processing parameters during the current data processing process. This association is stored in the historical process database as key-value pairs. The key is a unique identifier for the current data processing process, containing information such as equipment number, contact opening and closing time, and contact opening and closing type. The value is the standardized current and voltage data sequence and associated processing information. This ensures that the current record can be quickly retrieved through the unique identifier during data processing, providing new qualified data support for historical similar record retrieval and reference interference pattern construction. At the same time, it enables the continuous enrichment and optimization of the historical process database, improving the accuracy and efficiency of data processing.

[0058] The control command input signal is a signal data that can be recognized by the control system of the electrical control equipment and used to generate equipment control commands. It serves as the link between real-time data processing and actual equipment control. First, key parameters are extracted from the standardized current and voltage data sequence. These key parameters are core signal parameters that reflect the actual opening and closing state of the contacts and serve as the basis for generating control commands. They mainly include the voltage stability value and the current waveform slope. The voltage stability value is the average voltage amplitude of all sampling points in the stable closing stage of the contacts in the standardized current and voltage data sequence. It can characterize the voltage working state after the contacts are stably closed and reflect the contact reliability. The current waveform slope is the slope value obtained by linearly fitting the amplitude change trend of the current signal in the standardized current and voltage data sequence using the least squares method. The least squares method is a mathematical method that solves the data fitting line by minimizing the sum of squares of errors. Its core working principle is to calculate the minimum sum of squares of errors between the fitted line and the actual data points to determine the slope and intercept of the fitted line. The input data of this method is the current signal sampling points in the standardized current and voltage data sequence, and the output data is the fitted line of the current waveform and the corresponding slope value. The current waveform slope can characterize the rate of change of current during the opening and closing of the contacts and reflect the action state and process of the contacts. After extracting key parameters, the extracted key parameters such as voltage stability value and current waveform slope are associated with the control logic of the electrical control equipment to map the parameters to control command input signals. For example, the preset closure judgment threshold can be set to 90% of the equipment's rated voltage. That is, when the voltage stability value exceeds the preset closure judgment threshold, a control command input signal for contact closure is generated. When the current waveform slope shows a preset opening trend, the preset opening trend can be defined as: performing sliding window detection on the current waveform slope. If the average slope of three consecutive windows is less than -5A / s, it is determined that the contact is opening, and a control command input signal for contact opening is generated. The mapped control command input signal is directly transmitted to the control system of the electrical control equipment, providing data basis for the control system to generate accurate equipment control commands. Finally, the real-time data processing process of the entire electrical control equipment is completed, realizing a closed-loop process from signal acquisition and interference elimination to control command input signal generation.

[0059] The above process completes the time-domain and frequency-domain feature verification, parameter backtracking adjustment, signal standardization conversion, and generation of control command input signals for a complete and smooth signal sequence. The multi-dimensional feature verification method comprehensively ensures the quality of the output signal from both time and frequency dimensions, effectively preventing unqualified signals from entering the control command generation stage. Parameter backtracking adjustment when feature verification fails enables iterative optimization of signal processing, ensuring that the final output signal fully meets the usage requirements of the electrical control equipment. Signal standardization processing achieves precise adaptation with the control command generation system, while the generation of control command input signals effectively connects real-time data processing with actual equipment control. This entire process forms the final quality control and output stage of real-time data processing for electrical control equipment. Together with the preceding steps of signal acquisition, denoising, feature extraction, interference elimination, and residual analysis, it forms a complete closed-loop processing system, ensuring the accuracy, reliability, and practicality of the entire data processing process. This provides high-quality real-time data support for the stable operation and precise control of electrical control equipment, significantly improving the control accuracy and operational stability of the equipment.

[0060] The above describes a real-time data processing method for electrical control equipment according to an embodiment of this application. The following describes a real-time data processing system for electrical control equipment according to an embodiment of this application. Please refer to [link / reference]. Figure 4 One embodiment of a real-time data processing system for electrical control equipment in this application includes: The feature extraction unit is used to collect the current and voltage signals at the moment of opening and closing of the contacts of electrical control equipment through sensors to obtain a transient signal sequence. The transient signal sequence is subjected to preliminary denoising processing to generate a preliminary smooth signal. The temporal features are extracted from the preliminary smooth signal and normalized to generate a standardized temporal feature description vector.

[0061] The pattern construction unit is used to retrieve historical records similar to the time-series feature description vector from a pre-established historical process database, construct a reference interference pattern for the current data processing based on the historical records, compare the preliminary smoothed signal with the reference interference pattern for interference comparison and error correction, and then perform segmented differential filtering for different interference types to generate a complete smoothed signal sequence.

[0062] The signal analysis unit is used to perform residual analysis on complete smooth signal sequences and transient signal sequences, extract unique interference features, associate and store the unique interference features with the complete smooth signal sequence, and update the historical process database.

[0063] The real-time processing unit is used to verify the time-domain and frequency-domain features of the complete smooth signal sequence. After the verification is successful, the complete smooth signal sequence is converted into a standardized current and voltage data sequence. Real-time data processing is completed based on the standardized current and voltage data sequence. If the verification fails, it is backtracked to the preliminary denoising stage for reprocessing until the verification is successful.

[0064] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0065] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0066] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A real-time data processing method for electrical control equipment, characterized in that, The method includes: The current and voltage signals of electrical control equipment contacts are collected by sensors at the moment of opening and closing to obtain transient signal sequences. The transient signal sequences are subjected to preliminary denoising processing to generate preliminary smoothed signals. The temporal features of the preliminary smoothed signals are extracted and normalized to generate standardized temporal feature description vectors. Retrieve historical records similar to the time-series feature description vector from a pre-established historical process database, construct a reference interference pattern for the current data processing process based on the historical records, compare the preliminary smoothed signal with the reference interference pattern for interference comparison and error correction, and then perform segmented differential filtering for different interference types to generate a complete smoothed signal sequence. Residual analysis is performed on the complete smooth signal sequence and the transient signal sequence to extract unique interference features. The unique interference features are associated with and stored with the complete smooth signal sequence, and the historical process database is updated. The complete smooth signal sequence is verified in both the time and frequency domains. If the verification is successful, the complete smooth signal sequence is converted into a standardized current and voltage data sequence. Real-time data processing is then performed based on the standardized current and voltage data sequence. If the verification fails, the process is reverted to the initial denoising stage and reprocessed until the verification is successful.

2. The real-time data processing method for electrical control equipment according to claim 1, characterized in that, Generate a preliminary smoothed signal, including: The sensor synchronously acquires dual-channel current and voltage signals at the moment of contact opening and closing. For the spike pulses caused by contact jitter and the continuous oscillations caused by arcing in the dual-channel current and voltage signals, a wavelet threshold denoising method is used to process them to generate a preliminary smoothed transient signal sequence. If the transient signal sequence has local noise, the noise distribution characteristics are recorded. At the same time, it is determined whether the denoising result of the dual-channel current and voltage signals meets the preset standard. If it does not meet the standard, the denoising parameters are adjusted and reprocessed until the denoising effect meets the preset standard, and the final preliminary smoothed signal is generated.

3. The real-time data processing method for electrical control equipment according to claim 1, characterized in that, Generate standardized time-series feature description vectors, including: The preliminary smoothed signal is processed using a time-series segmentation method based on envelope detection to determine the time nodes at the start of the mechanical action of the contact, the initial contact, the stable closure, and the extinction of the arc. The amplitude distribution characteristics of the interference waveform envelope at the boundary points of each time node are extracted, and the regularity data of the repetition interval of the interference pulse are statistically analyzed. A time-series feature description vector is constructed based on the amplitude distribution characteristics and the regularity data. The data of each dimension of the time-series feature description vector are normalized to generate a standardized time-series feature description vector.

4. The real-time data processing method for electrical control equipment according to claim 1, characterized in that, Construct a reference disturbance pattern for the current data processing procedure, including: Based on the Euclidean distance between the time-series feature description vector and the feature vectors of each historical record in the historical process database, several sets of historical records with similarity higher than a preset similarity threshold are extracted; from these sets of historical records, the phase shift data relative to the contact action at the time of interference occurrence, the correlation between interference amplitude and voltage polarity, and the temporal coupling degree between interference and current zero crossing point are extracted; based on the phase shift data, the correlation relationship, and the temporal coupling degree, a reference interference mode adapted to the current data processing process is constructed.

5. The real-time data processing method for electrical control equipment according to claim 4, characterized in that, Constructing a reference disturbance pattern for the current data processing process also includes: The phase offset data, the relevant correspondence, and the temporal coupling degree are normalized respectively to generate corresponding normalized features; the weight values ​​corresponding to the normalized features are obtained based on the feature stability coefficients in historical records, and the weight values ​​are adaptively corrected in combination with the equipment operating conditions; the normalized features are constructed into a multi-dimensional feature vector, and the multi-dimensional feature vector is weighted based on the corrected weight values ​​to obtain the feature core values; the feature core values, the corrected weight values, and the historical interference type labels are combined to construct a basic reference interference pattern; Extract the real-time features of the current preliminary smoothing signal and normalize them to obtain a real-time feature vector. Calculate the deviation between the real-time feature vector and the feature vector of the basic reference interference mode. If the deviation exceeds a preset deviation threshold, expand the search history and construct a mode. Otherwise, split the basic reference interference mode into feature dimension layers according to the contact action stage and label the feature value range of each stage. Add adaptation tags to generate a reference interference mode that adapts to the current data processing process.

6. The real-time data processing method for electrical control equipment according to claim 1, characterized in that, The preliminary smoothed signal is compared with the reference interference mode for interference comparison and error correction, including: The preliminary smoothed signal is compared point by point with the reference interference pattern. If the amplitude distribution characteristic deviation of the interference waveform envelope of a local segment exceeds a preset range, a Kalman filter is activated to recursively estimate the state of the random disturbance dominated by touch jitter in the local segment, generating a preliminary estimated smoothed trajectory signal. A segmented deviation sequence is calculated between the smoothed trajectory signal and the reference interference pattern. If the segmented deviation sequence has abnormal deviations that exceed a preset length, the signal features of the deviation segment are extracted and binary classification is performed using a support vector machine. Based on the classification results, a reliable signal subsequence is marked.

7. The real-time data processing method for electrical control equipment according to claim 6, characterized in that, Generate a complete smooth signal sequence, including: Align the timing of the trusted signal subsequence with the timing of the interference stage in the reference interference mode to obtain the interference type marked in the reference interference mode. If the interference type is dominated by arc interference, an adaptive weighted average filtering process with the forgetting factor increasing with the interference intensity is applied to the corresponding signal stage. If the interference type is dominated by contact jitter, a median sorting depth filtering process positively correlated with the jitter pulse density is applied to the corresponding signal stage. The filtered signals of each stage are spliced ​​together according to the original timing and the splicing boundary is smoothed to generate a complete smooth signal sequence. At the same time, the smoothness of the complete smooth signal sequence is verified. If the verification fails, the filtering parameters are adjusted and the filtering is repeated until the verification passes.

8. The real-time data processing method for electrical control equipment according to claim 1, characterized in that, Residual analysis is performed on the complete smooth signal sequence and the transient signal sequence, including: The residual sequence is obtained by calculating the pointwise residual between the complete smooth signal sequence and the transient signal sequence. The residual sequence is then segmented by a threshold and the interference segments are identified. The duration characteristics of each interference segment are extracted and the segmented probability density of the interference duration is statistically obtained. Local peak points within each interference segment are identified and the attenuation rate characteristics of the local peaks of the interference are statistically obtained. The segmented probability density and the attenuation rate characteristics are used as the unique interference features and associated with the complete smooth signal sequence and stored in the historical process database. The reference interference patterns of the corresponding categories in the historical process database are then updated with weights based on the newly stored data, thus completing the update of the historical process database.

9. The real-time data processing method for electrical control equipment according to claim 1, characterized in that, The time-domain and frequency-domain features of the complete smooth signal sequence are verified, including: The average amplitude, peak value, duration, and time-domain features of the complete smooth signal sequence are calculated. A Fourier transform is performed on the complete smooth signal sequence to obtain the frequency domain spectrum. Based on the frequency domain spectrum, the dominant frequency component and noise band energy are extracted as frequency domain features. The time-domain features and the frequency domain features are compared with corresponding preset standards. If both meet the corresponding preset standards, the complete smooth signal sequence is subjected to data format conversion and sampling rate unification processing to generate a standardized current and voltage data sequence. At the same time, the standardized current and voltage data sequence is stored in the historical process database, and key parameters are extracted from the standardized current and voltage data sequence and mapped to control command input signals of electrical control equipment.

10. A real-time data processing system for electrical control equipment, used to implement the real-time data processing method for electrical control equipment as described in any one of claims 1-9, characterized in that, The system includes: The feature extraction unit is used to collect the current and voltage signals at the moment of opening and closing of the contacts of the electrical control equipment through the sensor to obtain the transient signal sequence, perform preliminary denoising processing on the transient signal sequence to generate a preliminary smooth signal, extract the time series features from the preliminary smooth signal and perform normalization processing to generate a standardized time series feature description vector. The pattern construction unit is used to retrieve historical records similar to the time-series feature description vector from a pre-established historical process database, construct a reference interference pattern for the current data processing process based on the historical records, perform interference comparison and error correction processing on the preliminary smoothed signal and the reference interference pattern, and then perform segmented differential filtering for different interference types to generate a complete smoothed signal sequence. The signal analysis unit is used to perform residual analysis on the complete smooth signal sequence and the transient signal sequence, extract unique interference features, associate and store the unique interference features with the complete smooth signal sequence, and update the historical process database. The real-time processing unit is used to perform time-domain and frequency-domain feature verification on the complete smooth signal sequence. After the verification is successful, the complete smooth signal sequence is converted into a standardized current-voltage data sequence. Real-time data processing is completed based on the standardized current-voltage data sequence. If the verification fails, the process is backtracked to the preliminary denoising stage for reprocessing until the verification is successful.

Images