A method and device for capturing and analyzing early sticking signs of a circuit breaker of a switchgear

Abstract

The present application relates to a kind of switchgear circuit breaker early jamming sign capture analysis method and device, it is related to switchgear abnormal detection technical field.The present application is filtered and pretreated to original current signal sequence using self-adapting wavelet threshold denoising to closing and opening coil current signal;Using first-order difference sequence to combine sliding window variance decision, the action stage of original current signal sequence is adaptively segmented and intercepted, and effective waveform section is obtained;For core no-load movement section and mechanical load work section, empirical mode decomposition and instantaneous frequency spectrum analysis are carried out, and high-frequency tremor energy mean and frequency standard deviation are extracted;Effective waveform section is extracted to curvature cumulative change and time length micro deviation rate;Based on the Mahalanobis distance between multidimensional feature vector and historical health baseline, jamming sensitivity index for quantifying early jamming degree is obtained;Linear regression trend prediction and contribution degree traceability are used to output graded early warning results and jamming site positioning information, and accurate fault early warning and positioning are realized.

Description

Technical Field

[0001] This invention relates to the field of switchgear anomaly detection technology, and in particular to a method and device for capturing and analyzing early signs of jamming in switchgear circuit breakers. Background Technology

[0002] Switchgear is a critical piece of equipment in power systems used for controlling, protecting, and isolating electrical devices. The reliability of its operating mechanism directly affects the safe and stable operation of the power grid. During long-term operation, circuit breaker operating mechanisms are prone to jamming due to factors such as wear of mechanical parts, aging of lubricating grease, fatigue deformation of springs, and eccentricity of transmission links. In its early stages, jamming often manifests as increased micro-frictional resistance during movement. If not detected and addressed promptly, it will gradually evolve into severe mechanical jamming, leading to circuit breaker failure to operate or malfunction, and ultimately, power grid accidents. Therefore, accurate detection and early warning of early jamming symptoms in switchgear operating mechanisms are crucial technical means to improve the operational reliability of power equipment.

[0003] Currently, the widely used methods for monitoring the status of switchgear operating mechanisms in engineering practice are mainly based on time-domain analysis of the current waveforms of the opening and closing coils. For example, Chinese patent CN202110160424.4 discloses a method for monitoring the status of circuit breaker operating mechanisms. This method extracts several fixed inflection points in the current waveform (such as the starting point of the coil current rise, the starting point of the iron core, the contact point, the current drop point, and other abrupt change points) for time-domain analysis, and judges the working status of the operating mechanism based on the time interval between each inflection point and the change in current amplitude. This method has the advantages of simple principle and convenient engineering implementation, and has achieved certain results in conventional fault detection.

[0004] However, the aforementioned existing technologies have the following technical defects: First, when the operating mechanism experiences early lubrication problems or slight eccentricity, the main peak value and total time of the time-domain waveform are often still within the normal range. However, the increased micro-friction resistance during its movement will generate high-frequency micro-oscillations superimposed on the power frequency current. The amplitude of these high-frequency micro-oscillations is usually in the milliampere level, much smaller than the amplitude of the power frequency current, making it difficult to capture using the time-domain threshold method with a fixed inflection point. Second, existing methods rely solely on time-domain features and lack the ability to mine the joint time-frequency domain features of the signal. They cannot distinguish the weak signs caused by abnormalities in different mechanical components and are difficult to accurately locate jammed parts. Third, existing methods mostly use a single threshold for fault determination and lack the ability to adaptively learn and track trends from the historical health baseline of the equipment. They are easily affected by single, accidental fluctuations, resulting in high false alarm and false negative rates. Fourth, existing methods fail to effectively integrate features from different physical dimensions, making it impossible to form a unified quantitative indicator and meet the quantitative assessment needs of operation and maintenance decisions.

[0005] In summary, existing technologies are insufficient to meet the practical needs of accurately capturing and locating early signs of jamming in switchgear operating mechanisms. There is an urgent need for a new monitoring method that can simultaneously mine time-domain and frequency-domain features, adaptively learn the device health baseline, and achieve multi-dimensional feature fusion and trend tracking. Summary of the Invention

[0006] In order to solve the above-mentioned technical problems, or at least partially solve the above-mentioned technical problems, and in view of the above-mentioned shortcomings, the present invention provides a method and device for capturing and analyzing early jamming symptoms of switchgear circuit breakers.

[0007] In a first aspect, the present invention provides a method for capturing and analyzing early signs of jamming in switchgear circuit breakers, comprising: Step 1: Collect the opening and closing coil current of the switchgear circuit breaker and preprocess it to obtain the original current signal sequence; Step 2: Adaptive segmentation and truncation of the circuit breaker operation stage are performed on the original current signal sequence using a first-order difference sequence combined with a sliding window variance action stage decision method to obtain the effective waveform segments of the operation stage, and to divide the circuit breaker core no-load movement stage and the mechanical load work stage. Step 3: Perform empirical mode decomposition on the core no-load motion segment and the mechanical load working segment in the effective waveform segment of the action, respectively, and select the intrinsic mode function components whose instantaneous frequency is within the preset high frequency band to form a high-frequency feature set; perform Hilbert transform on each component in the high-frequency feature set to obtain the analytical signal, and calculate the instantaneous frequency and envelope amplitude; calculate the mean of high-frequency flutter energy based on the envelope amplitude, and calculate the frequency standard deviation based on the instantaneous frequency; Step 4: Perform amplitude normalization and smoothness distortion calculation on the effective waveform segment of the action to extract the cumulative curvature change, and calculate the duration micro-deviation rate of the effective waveform segment of the action; the smoothness distortion calculation adopts the second derivative curvature accumulation algorithm. Step 5: Construct a multi-dimensional feature vector of the effective waveform segment signal of the circuit breaker operation and a historical health baseline. Based on the Mahalanobis distance between the multi-dimensional feature vector and the historical health baseline, obtain the jamming sensitivity index used to quantify the early jamming degree. The multi-dimensional feature vector includes: mean high-frequency flutter energy, frequency standard deviation, cumulative curvature change, and duration micro-deviation rate. Step 6: Use linear regression trend prediction to track and analyze the sluggishness index sequence. Through multidimensional feature vector contribution analysis, output the graded early warning results and the corresponding sluggishness location information.

[0008] Furthermore, an adaptive wavelet thresholding algorithm is used to preprocess the original opening and closing coil current signal by independently estimating the noise level layer by layer based on the distribution of detail coefficients to calculate the threshold, and then using the threshold to process the detail coefficients through soft thresholding for wavelet reconstruction. This preprocessing includes: The Daubechies wavelet basis db4 is selected to perform J-level orthogonal wavelet decomposition on the noisy discrete current sequence X[n], and the scaling function, wavelet function, approximation coefficient and detail coefficient of each level are obtained. The noise level is independently estimated layer by layer based on the distribution of detail coefficients to calculate the threshold. The median absolute deviation is used to estimate the noise standard deviation of the detail coefficients at each layer, and the threshold for each layer is calculated based on these noise standard deviation estimates. In the formula, Let N be the noise standard deviation estimate of the detail coefficients of the j-th layer, and N be the total number of samples. Based on the obtained thresholds for each layer The detail coefficients are processed using a soft threshold shrinkage function: ; In the formula, sgn(·) is the sign function; Wavelet reconstruction is performed using the processed detail coefficients and the unprocessed approximation coefficients to obtain the original current signal sequence.

[0009] Furthermore, the decision-making process for the first-order difference sequence combined with the sliding window variance action phase includes: Perform a first-order difference operation on the original current signal sequence to obtain a difference sequence; Slide a sliding window of a set length along the difference sequence and calculate the variance of the difference sequence within the sliding window; Set the start and end thresholds; When the variance within the sliding window first exceeds the start threshold and remains there for at least the shortest duration, it is determined as the start time of the action; when the variance within the sliding window falls from the peak to below the end threshold and remains there, it is determined as the end time of the action; the waveform segment between the start time and the end time of the action is extracted as the effective waveform segment of the action, and the effective waveform segment of the action is divided into the circuit breaker core no-load movement segment and the mechanical load work segment.

[0010] Furthermore, step three includes: In each iteration, cubic spline interpolation is used to fit the local maximum and local minimum envelopes of the signals for the current unloaded core movement segment and the mechanically loaded working segment, respectively, and the mean of the upper and lower envelopes is calculated. The mean of the envelopes is subtracted from the current signal to obtain new candidate components. The above iterative process is repeated until the stopping criterion is met to obtain the q-th order intrinsic mode function component. The residual component from the previous step is subtracted from the current intrinsic mode function component to update the residual component. The signals for the unloaded core movement segment and the mechanically loaded working segment are reconstructed using the residual components of each order and the intrinsic mode function components. The intrinsic mode function components whose instantaneous frequency falls within a preset high-frequency band are selected to form a high-frequency feature set; Hilbert transform is performed on each intrinsic mode function component in the high-frequency feature set to obtain analytic signals of each order. The instantaneous frequency and envelope amplitude of each order are calculated using the analytic signals and intrinsic mode function components. The mean high-frequency flutter energy is calculated based on the envelope amplitude of each order during the unloaded movement segment and the mechanically loaded working segment of the iron core, and the standard deviation of the frequency is calculated based on the instantaneous frequency of each order during the unloaded movement segment and the mechanically loaded working segment of the iron core.

[0011] Furthermore, step four includes: Amplitude normalization is performed on the effective waveform segments of the action; The smoothness distortion reduction calculation for the normalized effective motion waveform segment includes: calculating the second derivative of the normalized effective motion waveform segment signal. And accumulate the second derivative during the duration of the effective waveform segment of the action. The cumulative curvature change of the effective waveform segment of the entire motion is obtained. ; Extract the actual total duration of the valid waveform segment of the current operation of the circuit breaker. and compared with the historical average standard duration of the circuit breaker. Calculation of duration micro-deviation rate: .

[0012] Furthermore, a healthy sample matrix is ​​constructed by calling the multi-dimensional feature vectors of the valid waveform segments corresponding to the normal actions of the circuit breaker after a set number of pre-set actions, and the mean vector is calculated. Covariance Matrix As a historical health baseline; the standardized distance between the current feature vector and the historical health baseline is calculated using Mahalanobis distance as the Caesars sensitivity index (CSI). .

[0013] Furthermore, the tracking and analysis of the Caesars sensitivity index series using linear regression trend prediction includes: The sticking sensitivity index, calculated based on the effective waveform segment signal of each action, is recorded in order of the number of actions to form a sticking sensitivity index-number of actions trend sequence; Linear regression was used to fit the stickiness sensitivity index of the most recent actions to obtain the recent growth slope. When the recent growth slope β is confirmed to be greater than the set slope threshold twice in a row, and the jamming sensitivity index CSI is between the light alarm threshold and the heavy alarm threshold, a first-level lubrication maintenance recommendation is output. When the sensitivity index (CSI) is greater than the critical alarm threshold, a secondary mechanism maintenance alarm is output.

[0014] Furthermore, the source analysis of the contribution of multidimensional feature vectors includes: After identifying the anomaly, backtrack the contribution of each dimension of the current anomaly's multidimensional feature vector; If the sum of the deviation contributions of the mean energy of high-frequency vibration and the standard deviation of frequency is the largest, the jamming is determined to originate from the sliding friction pair of the iron core or push rod; if the sum of the deviation contributions of the cumulative change in curvature and the micro deviation rate of duration is the largest, the jamming is determined to originate from the nonlinear jamming of the transmission link or spring mechanism.

[0015] Furthermore, the contribution of the i-th feature of the multidimensional feature vector to the Mahalanobis distance. Determined by the product of the deviation values ​​of each dimension's features and the inverse covariance matrix:

[0016] in, Multidimensional feature vectors The i-th and j-th elements, Multidimensional feature vectors The mean vector, Here is the covariance matrix, and CSI is the Cauchy sensitivity index.

[0017] Secondly, the present invention provides a device for capturing and analyzing early signs of jamming in switchgear circuit breakers, comprising: at least one processing unit, wherein the processing unit is connected to a storage unit and a data acquisition unit via a bus unit, and the processing unit implements the aforementioned method for capturing and analyzing early signs of jamming in switchgear circuit breakers by running a computer-executable program stored in the storage unit.

[0018] The technical solutions provided in the embodiments of the present invention have the following advantages compared with the prior art: This application employs an adaptive wavelet threshold denoising algorithm that independently estimates the noise level layer by layer based on the distribution of detail coefficients to calculate the threshold. Based on the threshold, the detail coefficients are processed by soft threshold shrinkage and then used for wavelet reconstruction. This algorithm performs filtering preprocessing on the original opening and closing coil current signal, which can more accurately filter out high-frequency white noise and external power frequency harmonic interference, while retaining the weak high-frequency jitter features in the signal, providing a high signal-to-noise ratio signal basis for subsequent feature extraction.

[0019] This invention uses a first-order difference sequence combined with a sliding window variance decision to adaptively segment the original current signal sequence for the action phase. It can automatically identify the start and end times of the action without manually setting a fixed time window, adapting to the differences in action duration under different circuit breaker models and operating conditions, and improving the accuracy and versatility of extracting the effective waveform segment of the action.

[0020] By performing empirical mode decomposition on the unloaded movement section and the mechanical load working section of the iron core, the high-frequency micro-oscillation characteristics superimposed on the power frequency current can be extracted: the mean energy and standard deviation of the high-frequency vibration. This makes up for the shortcomings of existing technologies that rely only on time-domain inflection point analysis, and achieves accurate capture of early weak jamming symptoms.

[0021] This invention employs a second-derivative curvature accumulation algorithm to perform smoothness distortion quantification calculation, extracts the cumulative curvature change, and calculates the duration micro-deviation rate of the action. It quantifies the degree of jamming from the macroscopic time-domain deformation dimension, complementing the high-frequency micro-oscillation characteristics, and realizes multi-physical dimension joint analysis of the effective waveform segment of the same action.

[0022] This invention employs a Mahalanobis distance multidimensional fusion algorithm to compress heterogeneous features into a dimensionless Caesar sensitivity index (CSI). It utilizes the covariance matrix to adaptively weight and normalize features of different dimensions, avoiding the subjectivity of manually setting weights and providing a unified benchmark for the quantitative evaluation of Caesar sensitivity.

[0023] This invention employs linear regression trend prediction to track and analyze the jamming sensitivity index sequence. Through multidimensional feature vector contribution analysis, it outputs graded early warning results and corresponding jamming location information. The recent growth slope is used to determine the development trend of jamming, eliminating the interference of single accidental fluctuations. Simultaneously, the internal projection distribution of high-dimensional feature vectors is used to trace back to specific mechanical weak points, realizing a closed loop from data collection to operation and maintenance decision-making, providing operation and maintenance personnel with accurate fault early warning and location information.

[0024] In summary, this invention effectively solves the technical problems of existing technologies, such as difficulty in capturing early and weak signs of jamming, difficulty in locating jamming locations, and susceptibility to interference from accidental fluctuations, through the synergistic combination of time-frequency domain joint feature extraction, multi-dimensional fusion, and trend tracking. This significantly improves the accuracy and reliability of status monitoring of switchgear operating mechanisms. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A flowchart of a method for capturing and analyzing early signs of jamming in switchgear circuit breakers provided in an embodiment of the present invention; Figure 2 A flowchart of the adaptive wavelet threshold denoising algorithm provided in an embodiment of the present invention; Figure 3 The flowchart provided in this embodiment of the invention describes an adaptive segmentation and truncation of the original current signal sequence during the action phase, using a first-order difference sequence combined with a sliding window variance decision, to obtain the effective waveform segment of the action. Figure 4 A flowchart for calculating the mean energy and standard deviation of high-frequency tremor provided in an embodiment of the present invention; Figure 5 The flowchart provided in this embodiment of the invention extracts a multidimensional feature vector from the effective waveform segment of the circuit breaker operation and the historical health baseline of the circuit breaker, and obtains a jamming sensitivity index for quantifying the early jamming degree based on the Mahalanobis distance between the multidimensional feature vector and the historical health baseline of the circuit breaker. Figure 6 A schematic diagram of the device for capturing and analyzing early signs of jamming in switchgear circuit breakers provided in an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0030] Example 1: like Figure 1 As shown in the figure, this embodiment provides a method for capturing and analyzing early signs of jamming in switchgear circuit breakers, which includes the following steps: Step 1: Collect the opening and closing coil current of the switchgear circuit breaker and preprocess it to obtain the original current signal sequence I[n].

[0031] In this embodiment, a Hall current sensor is used to synchronously acquire the trip coil current signal of a 10kV vacuum circuit breaker at a sampling rate of 20kHz. and closing coil current signal The total number of sampling points for a single action is N=4000. The acquired noisy discrete current sequence X[n] contains both the power frequency current component reflecting the motion state of the operating mechanism and high-frequency white noise and external power frequency harmonic interference. To improve the accuracy of subsequent feature extraction, adaptive wavelet threshold denoising processing is required for the discrete current sequence X[n].

[0032] like Figure 2 As shown, an adaptive wavelet threshold denoising algorithm is used to preprocess the original opening and closing coil current signal by independently estimating the noise level layer by layer based on the distribution of detail coefficients, and then using soft thresholding to shrink the detail coefficients based on the threshold for wavelet reconstruction. This preprocessing includes: firstly, selecting the Daubechies wavelet basis db4, and performing J=5 layers of orthogonal wavelet decomposition on the noisy discrete current sequence X[n]. The decomposition formula is shown below: ; In the formula, j is the decomposition layer number, j=1,2,…,J; k is the index number of the wavelet coefficients of the j-th layer; For the J-th layer scaling function (approximate component basis function); Let be the wavelet function of the j-th layer (basic function of detail components); For the approximate coefficients (low-frequency components) of the Jth layer; is the detail coefficient (high-frequency component) of the j-th layer.

[0033] The noise level is independently estimated layer by layer based on the distribution of detail coefficients to calculate the threshold. The process includes: The noise standard deviation of the detail coefficients at each layer is estimated using the median absolute deviation. For the j-th layer, the noise standard deviation of the detail coefficients is estimated as follows: for: ; In the formula The total number of detail coefficients at level j is denoted as , and median(·) is the median operator.

[0034] The threshold for each layer is calculated based on the noise standard deviation estimate of the detail coefficients: .

[0035] Then, based on the obtained threshold values ​​for each layer... The detail coefficients are processed using a soft threshold shrinkage function: ; In the formula, sgn(·) is the sign function. Soft thresholding has better continuity than hard thresholding and can avoid the pseudo-Gibbs phenomenon during signal reconstruction.

[0036] Finally, using the processed detail coefficients and unprocessed approximation coefficients Wavelet reconstruction is performed to obtain the original current signal sequence I[n]: .

[0037] Step 2: Adaptive segmentation and truncation of the original current signal sequence for the action phase is performed using a first-order difference sequence combined with a sliding window variance decision to obtain the effective waveform segment of the action.

[0038] like Figure 3 As shown, the specific implementation process of this step includes: First, a first-order difference operation is performed on the original current signal sequence to obtain the difference sequence. : Differential operations can highlight the abrupt changes in a signal, making it easier to identify the start and end times of an action.

[0039] A sliding window of a set length is used to slide along the difference sequence, and the variance of the difference sequence within the sliding window is calculated. First, the mean of the difference sequence within the sliding window is calculated, expressed as: , Where M is the length of the sliding window, and in the example, the length M of the sliding window is 100; The variance of the difference sequence within the sliding window, calculated using the mean, is expressed as: ; The variance of a sliding window can reflect the degree of fluctuation of a signal within that window.

[0040] Set startup threshold and the end threshold .

[0041] The starting and ending times of an action are analyzed by comparing the variance within the sliding window with preset start and end thresholds. Specifically, the action begins when the variance within the sliding window first exceeds the start threshold. The action is considered to begin when the duration is not less than the shortest duration; when the variance within the sliding window falls from its peak to below the end threshold... If the action is maintained continuously, it is determined to be the end time t3 of the action.

[0042] The waveform segment between the start time t0 and the end time t3 of the action is selected as the valid waveform segment of the action.

[0043] The effective waveform segment of the operation is divided into the no-load movement segment of the circuit breaker core. and mechanical load working section ; In the formula, This represents the number of sampling points for the no-load moving section of the circuit breaker core. This represents the number of sampling points for the mechanical load operation segment. The no-load movement segment of the iron core corresponds to the period from the start of the iron core's movement to the contact point, while the mechanical load operation segment corresponds to the period from contact point to the end of the action. The two waveforms reflect the motion states of different components of the operating mechanism and require separate time-frequency analysis.

[0044] Step 3: Perform empirical mode decomposition on the core no-load motion segment and the mechanical load working segment in the effective waveform segment of the action, respectively, and select the intrinsic mode function components whose instantaneous frequency is within the preset high frequency band to form a high frequency feature set; perform Hilbert transform on each component in the high frequency feature set to obtain the analytical signal, and calculate the instantaneous frequency and envelope amplitude; calculate the mean high frequency flutter energy based on the envelope amplitude, and calculate the frequency standard deviation based on the instantaneous frequency.

[0045] like Figure 4 As shown, the specific implementation process of step three is as follows: First, empirical mode decomposition is performed on the core no-load movement segment and the mechanical load operation segment in the effective waveform segment of the circuit breaker operation. Inherent mode function components are extracted through an iterative screening process, including: For the unloaded movement segment of the iron core, in each iteration, cubic spline interpolation is used to fit the local maximum envelope of the current signal. and local minimum envelope Calculate the mean of the upper and lower envelopes. Subtract the envelope mean from the current signal to obtain new candidate components: Repeat the above iterative process until the stopping criterion (such as the standard deviation criterion or the S-number criterion) is met, and obtain the q-th order intrinsic mode function components: The residual components obtained from the decomposition are denoted as... : ; For the mechanical load operation segment, in each iteration, cubic spline interpolation is used to fit the local maximum envelope of the current signal. and local minimum envelope Calculate the mean of the upper and lower envelopes. Subtract the envelope mean from the current signal to obtain new candidate components: Repeat the above iterative process until the stopping criterion (such as the standard deviation criterion or the S-number criterion) is met, and obtain the q-th order intrinsic mode function components: The residual components obtained from the decomposition are denoted as... : .

[0046] Finally, the signals of the unloaded motion segment and the mechanical load working segment of the iron core are reconstructed using the residual components of each order and the intrinsic mode function components: , , In the formula, This represents the total number of intrinsic mode function components obtained from the decomposition of the unloaded motion segment of the iron core. The total number of intrinsic mode function components obtained by decomposing the work segment for the mechanical load.

[0047] The intrinsic mode function components whose instantaneous frequency falls within a preset high-frequency band are selected to form a high-frequency feature set. Specifically, for the unloaded movement segment of the iron core, the high-frequency feature set is as follows: ; For the mechanical load operation segment, the high-frequency characteristic set is as follows: .

[0048] Next, Hilbert transform is performed on each intrinsic mode function component in the high-frequency feature set to obtain analytic signals of each order. The instantaneous frequency and envelope amplitude of each order are calculated using the analytic signals and intrinsic mode function components.

[0049] Taking the unloaded movement segment of the iron core as an example, the Hilbert transform of the q-th order natural mode function component of the unloaded movement segment of the iron core is expressed as: ; in, Represents the Hilbert transform operator; Calculate the instantaneous frequencies of each order using the analytical signals and natural mode function components of the unloaded movement segment of the iron core: ; Calculate the envelope amplitude of each order using the analytical signals and intrinsic mode function components of the unloaded movement segment of the iron core: .

[0050] The calculation process for the instantaneous frequency and envelope amplitude of the mechanical load working segment is similar and will not be repeated here.

[0051] The average energy of high-frequency flutter was calculated based on the envelope amplitudes of each order during the unloaded movement and mechanically loaded working segments of the iron core. : , In the formula, These represent the total number of sampling points for the two high-frequency sets, respectively. . The envelope amplitudes of each order are calculated for the analytical signals and intrinsic mode function components of each order in the mechanical load working segment.

[0052] The average frequency is calculated based on the instantaneous frequencies of each order during the unloaded movement and mechanically loaded working sections of the iron core. : ; The standard deviation of the frequency is calculated based on the instantaneous frequencies and frequency mean of each order during the unloaded movement and mechanically loaded working sections of the iron core. : ; Average energy of high-frequency tremor It reflects the overall intensity and frequency standard deviation of high-frequency micro-oscillations during the action. It reflects the dispersion of the high-frequency micro-oscillation frequency, and together they characterize the micro-friction resistance characteristics during the motion of the operating mechanism.

[0053] Step 4: Perform amplitude normalization and smoothness distortion calculations on the effective waveform segment of the action to extract the cumulative curvature change, and calculate the duration micro-deviation rate of the effective waveform segment of the action; the smoothness distortion calculation adopts the second derivative curvature accumulation algorithm.

[0054] The specific implementation process of step four is as follows: First, the effective waveform segments of the action are normalized to map the waveform amplitude to the [0,1] interval, thus eliminating the influence of amplitude differences between different actions on subsequent calculations.

[0055] Then, smoothness distortion reduction calculation is performed on the normalized effective motion waveform segment. First, the second derivative of the normalized effective motion waveform segment signal is calculated. And accumulate the second derivative during the duration of the effective waveform segment of the action. The cumulative curvature change of the effective waveform segment of the entire motion is obtained. : Cumulative change in curvature This reflects the overall smoothness of the waveform. When the operating mechanism jams, the waveform will exhibit localized jitter and irregular abrupt changes, leading to... Increase.

[0056] Next, extract the actual total duration of the effective waveform segment of the current circuit breaker operation. and compared with the historical average standard duration of the circuit breaker. Calculation of duration micro-deviation rate Duration micro-deviation rate This reflects the relative change in motion duration. When the operating mechanism jams, the motion duration usually increases, leading to... It is a positive value.

[0057] This step and step three share the same effective waveform segment as the analysis object. However, step three focuses on the high-frequency vibration characteristics of the unloaded movement segment and the mechanically loaded work segment within this segment, while this step quantifies the macroscopic time-domain deformation of the entire effective waveform segment. Both analyze the same object from different physical dimensions, extracting the cumulative change in curvature. and duration micro-deviation rate The mean energy and standard deviation of the frequency tremor are input together with the data in step five for multidimensional fusion.

[0058] Step 5: Extract multidimensional feature vectors and historical health baselines of the circuit breaker from the effective waveform segments of the circuit breaker operation. Based on the Mahalanobis distance between the multidimensional feature vectors and the historical health baselines of the circuit breaker, obtain the jamming sensitivity index (CSI) to quantify the early jamming degree.

[0059] like Figure 5 As shown, the specific implementation process of step five is as follows: First, the mean value of high-frequency vibration energy, standard deviation of frequency, cumulative change of curvature, and cumulative change of curvature extracted from the effective waveform segment of the same action are combined into a multi-dimensional feature vector F.

[0060] Then, the multi-dimensional feature vector of the effective waveform segment corresponding to the pre-set normal action of the circuit breaker is used to construct a healthy sample matrix, and the mean vector is calculated. The covariance matrix C is used as the historical health baseline.

[0061] Finally, Mahalanobis distance is used to calculate the standardized distance between the current feature vector and the historical health baseline, which is then used as the Caesar Sensitivity Index (CSI). .

[0062] Mahalanobis distance adaptively weights and normalizes features of different dimensions using the inverse of the covariance matrix C, eliminating the influence of differences in feature dimensions on distance calculation. It also considers the correlation between features, enabling it to more accurately reflect the degree to which the current feature vector deviates from the healthy baseline. Therefore, the Caesar Sensitivity Index (CSI) can serve as a unified benchmark for quantifying the early degree of caking. By using Mahalanobis distance to adaptively weight and normalize features of different dimensions, heterogeneous features extracted from the same effective waveform segment of an action are compressed into a dimensionless index, providing a unique quantitative benchmark for tracking judgment.

[0063] Step Six: Use linear regression trend prediction to track and analyze the sluggishness sensitivity index sequence. Through multidimensional feature vector contribution analysis, output the graded early warning results and the corresponding sluggishness location information.

[0064] The specific implementation process of this step is as follows: First, the sticking sensitivity index calculated based on the effective waveform segment of each action is recorded in order of the number of actions, forming a sticking sensitivity index-number of actions trend sequence.

[0065] Then, linear regression is used to fit the jamming sensitivity index of the most recent actions to obtain the recent growth slope β. When the recent growth slope β is confirmed to be greater than the set slope threshold twice consecutively, and the jamming sensitivity index CSI is between the minor alarm threshold and the major alarm threshold, a first-level lubrication maintenance suggestion is output; where the slope threshold is 0.15, the minor alarm threshold is 3.0, and the major alarm threshold is 6.0; when the jamming sensitivity index CSI is greater than the major alarm threshold, a second-level mechanism maintenance alarm is output. By judging the development trend of jamming through the recent growth slope, the interference of single random fluctuations on the judgment is eliminated.

[0066] Multidimensional eigenvector contribution tracing analysis includes: After identifying an anomaly, backtrack the contribution of each dimension of the current anomaly's multidimensional feature vector; the contribution of the i-th feature of the multidimensional feature vector to the Mahalanobis distance. Determined by the product of the deviation values ​​of each dimension's features and the inverse covariance matrix: ; in, Multidimensional feature vectors The i-th and j-th elements, Multidimensional feature vectors The mean vector, Here is the covariance matrix, and CSI is the Cauchy sensitivity index.

[0067] If the sum of the deviation contributions of the mean energy of high-frequency vibration and the standard deviation of frequency is the largest, the jamming is determined to originate from the sliding friction pair of the iron core or push rod; if the sum of the deviation contributions of the cumulative change in curvature and the micro deviation rate of duration is the largest, the jamming is determined to originate from the nonlinear jamming of the transmission link or spring mechanism.

[0068] This step extends the single index output from step five to the time dimension for trend tracking, eliminating the interference of single, accidental fluctuations on the judgment; at the same time, it uses the internal projection distribution of high-dimensional feature vectors to trace back to specific mechanical weak links, completing the closed loop from data collection to operation and maintenance decision-making.

[0069] To verify the effectiveness of the method of the present invention, a 10kV vacuum circuit breaker was subjected to a 12-month field application verification at a 110kV substation. During the verification period, the circuit breaker performed a total of 86 opening and closing operations, of which the first 30 operations were used to construct a health sample matrix, and the subsequent 56 operations were used for jamming symptom detection and early warning verification.

[0070] During the 58th operation in the verification period, the jamming sensitivity index (CSI) output by the method of this invention was 3.42, with a recent growth slope β of 0.18. After two consecutive confirmations, a first-level lubrication and maintenance recommendation was output. Simultaneously, the contribution analysis showed that the sum of the deviations in the mean and standard deviation of high-frequency vibration energy accounted for 68%, indicating that the jamming originated from the sliding friction pair of the core or push rod. Based on the warning information, maintenance personnel performed targeted lubrication maintenance on the circuit breaker. After maintenance, the jamming sensitivity index (CSI) dropped to 1.85 during the 60th operation, returning to normal levels.

[0071] During the 78th operation in the verification period, the jamming sensitivity index (CSI) output by the method of this invention was 6.31, exceeding the critical alarm threshold, and a secondary mechanism maintenance alarm was output. Simultaneously, the contribution traceability results showed that the sum of the deviation contributions of the cumulative curvature change and the duration micro-deviation rate accounted for 72%, indicating that the jamming originated from nonlinear jamming of the transmission link or spring mechanism. Based on the warning information, maintenance personnel disassembled and inspected the circuit breaker, finding slight eccentric wear on the transmission link pin, consistent with the contribution traceability results. After maintenance, the jamming sensitivity index (CSI) dropped to 1.92 during the 80th operation, returning to normal levels.

[0072] Comparative verification results show that when the aforementioned jamming symptoms appeared, the time-domain characteristics (main peak value, total time) output by the existing technology (time-domain threshold method based on fixed inflection points) were within the normal range, failing to issue an early warning. However, the method of this invention can capture early jamming symptoms 2 to 4 operations in advance and accurately determine the jamming location, providing maintenance personnel with a valuable maintenance window and preventing further deterioration of the jamming fault. Statistics show that during the 12-month verification period, the early warning accuracy rate of the method of this invention reached 95.3%, the false alarm rate was only 2.3%, and the missed alarm rate was 0%, significantly better than the early warning accuracy rate (approximately 72.1%) and false alarm rate (approximately 18.6%) of the existing technology. The verification results demonstrate that the method of this invention can effectively capture early jamming symptoms of the switchgear operating mechanism, achieving accurate fault early warning and location, and has significant engineering application value.

[0073] Example 2: See Figure 6As shown, this embodiment of the invention provides a device for capturing and analyzing early signs of jamming in switchgear circuit breakers, comprising: at least one processing unit, the processing unit being connected to a storage unit and a data acquisition unit via a bus unit, the storage unit serving as a computer-readable storage medium, and capable of storing software programs, computer-executable programs, and modules, such as the software program, computer-executable program, and module corresponding to the method for capturing and analyzing early signs of jamming in switchgear circuit breakers in this embodiment of the invention. The processing unit implements the aforementioned method for capturing and analyzing early signs of jamming in switchgear circuit breakers by running the software program, computer-executable program, and module stored in the storage unit.

[0074] Of course, the computer program stored in the storage unit of the early jamming symptoms capture and analysis system for switchgear circuit breakers provided in the embodiments of the present invention is not limited to the operation of the method described above, and can also execute related operations in the early jamming symptoms capture and analysis method for switchgear circuit breakers provided in any embodiment of the present invention.

[0075] Example 3 This invention provides a computer-readable storage medium storing a computer program. When the computer program is executed, it implements the method for capturing and analyzing early jamming symptoms of switchgear circuit breakers.

[0076] The computer-readable storage medium provided in this embodiment of the invention stores a computer program that is not limited to the operation of the method described above, but can also execute related operations in the method for capturing and analyzing early jamming symptoms of switchgear circuit breakers provided in any embodiment of the invention.

[0077] In the embodiments provided by this invention, it should be understood that the disclosed structures and methods can be implemented in other ways. For example, the structural embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, structures, or units, and may be electrical, mechanical, or other forms.

[0078] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0080] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for capturing and analyzing early signs of jamming in switchgear circuit breakers, characterized in that, include: Step 1: Collect the opening and closing coil current of the switchgear circuit breaker and preprocess it to obtain the original current signal sequence; Step 2: Adaptive segmentation and truncation of the circuit breaker operation stage are performed on the original current signal sequence using a first-order difference sequence combined with a sliding window variance action stage decision method to obtain the effective waveform segments of the operation stage, and to divide the circuit breaker core no-load movement stage and the mechanical load work stage. Step 3: Perform empirical mode decomposition on the core no-load motion segment and the mechanical load working segment in the effective waveform segment of the action, respectively, and select the intrinsic mode function components whose instantaneous frequency is within the preset high frequency band to form a high-frequency feature set; perform Hilbert transform on each component in the high-frequency feature set to obtain the analytical signal, and calculate the instantaneous frequency and envelope amplitude; calculate the mean of high-frequency flutter energy based on the envelope amplitude, and calculate the frequency standard deviation based on the instantaneous frequency; Step 4: Perform amplitude normalization and smoothness distortion calculation on the effective waveform segment of the action to extract the cumulative curvature change, and calculate the duration micro-deviation rate of the effective waveform segment of the action; the smoothness distortion calculation adopts the second derivative curvature accumulation algorithm. Step 5: Construct a multi-dimensional feature vector of the effective waveform segment signal of the circuit breaker operation and a historical health baseline. Based on the Mahalanobis distance between the multi-dimensional feature vector and the historical health baseline, obtain the jamming sensitivity index used to quantify the early jamming degree. The multi-dimensional feature vector includes: mean high-frequency flutter energy, frequency standard deviation, cumulative curvature change, and duration micro-deviation rate. Step 6: Use linear regression trend prediction to track and analyze the sluggishness index sequence. Through multidimensional feature vector contribution analysis, output the graded early warning results and the corresponding sluggishness location information.

2. The method for capturing and analyzing early jamming symptoms of switchgear circuit breakers according to claim 1, characterized in that, An adaptive wavelet thresholding algorithm is used to preprocess the original opening and closing coil current signal by independently estimating the noise level layer by layer based on the distribution of detail coefficients to calculate the threshold, and then processing the detail coefficients through soft thresholding based on the threshold for wavelet reconstruction. This preprocessing includes: The Daubechies wavelet basis db4 is selected to perform J-level orthogonal wavelet decomposition on the noisy discrete current sequence X[n], and the scaling function, wavelet function, approximation coefficient and detail coefficient of each level are obtained. The noise level is independently estimated layer by layer based on the distribution of detail coefficients to calculate the threshold. The median absolute deviation is used to estimate the noise standard deviation of the detail coefficients at each layer, and the threshold for each layer is calculated based on these noise standard deviation estimates. In the formula, Let N be the noise standard deviation estimate of the detail coefficients of the j-th layer, and N be the total number of samples. Based on the obtained thresholds for each layer The detail coefficients are processed using a soft threshold shrinkage function: ； In the formula, sgn(·) is the sign function; Wavelet reconstruction is performed using the processed detail coefficients and the unprocessed approximation coefficients to obtain the original current signal sequence.

3. The method for capturing and analyzing early signs of jamming in switchgear circuit breakers according to claim 1, characterized in that, The first-order difference sequence combined with the sliding window variance action phase decision includes: Perform a first-order difference operation on the original current signal sequence to obtain a difference sequence; Slide a sliding window of a set length along the difference sequence and calculate the variance of the difference sequence within the sliding window; Set the start and end thresholds; When the variance within the sliding window first exceeds the start threshold and remains there for at least the shortest duration, it is determined as the start time of the action; when the variance within the sliding window falls from the peak to below the end threshold and remains there, it is determined as the end time of the action; the waveform segment between the start time and the end time of the action is extracted as the effective waveform segment of the action, and the effective waveform segment of the action is divided into the circuit breaker core no-load movement segment and the mechanical load work segment.

4. The method for capturing and analyzing early signs of jamming in switchgear circuit breakers according to claim 1, characterized in that, Step three includes: In each iteration, cubic spline interpolation is used to fit the local maximum and local minimum envelopes of the signals for the current unloaded core movement segment and the mechanically loaded working segment, respectively, and the mean of the upper and lower envelopes is calculated. The mean of the envelopes is subtracted from the current signal to obtain new candidate components. The above iterative process is repeated until the stopping criterion is met to obtain the q-th order intrinsic mode function component. The residual component from the previous step is subtracted from the current intrinsic mode function component to update the residual component. The signals for the unloaded core movement segment and the mechanically loaded working segment are reconstructed using the residual components of each order and the intrinsic mode function components. The intrinsic mode function components whose instantaneous frequency falls within a preset high-frequency band are selected to form a high-frequency feature set; Hilbert transform is performed on each intrinsic mode function component in the high-frequency feature set to obtain analytic signals of each order. The instantaneous frequency and envelope amplitude of each order are calculated using the analytic signals and intrinsic mode function components. The mean high-frequency flutter energy is calculated based on the envelope amplitude of each order during the unloaded movement segment and the mechanically loaded working segment of the iron core, and the standard deviation of the frequency is calculated based on the instantaneous frequency of each order during the unloaded movement segment and the mechanically loaded working segment of the iron core.

5. The method for capturing and analyzing early jamming symptoms of switchgear circuit breakers according to claim 1, characterized in that, Step four includes: Amplitude normalization is performed on the effective waveform segments of the action; The smoothness distortion reduction calculation for the normalized effective motion waveform segment includes: calculating the second derivative of the normalized effective motion waveform segment signal. And accumulate the second derivative during the duration of the effective waveform segment of the action. The cumulative curvature change of the effective waveform segment of the entire motion is obtained. ; Extract the actual total duration of the valid waveform segment of the current operation of the circuit breaker. and compared with the historical average standard duration of the circuit breaker. Calculation of duration micro-deviation rate: 。 6. The method for capturing and analyzing early jamming symptoms of switchgear circuit breakers according to claim 1, characterized in that, Before calling the circuit breaker a set number of times, a health sample matrix is ​​constructed from the multidimensional feature vectors of the valid waveform segments corresponding to the normal actions of the actions, and the mean vector is calculated. Covariance Matrix As a historical health baseline; the standardized distance between the current feature vector and the historical health baseline is calculated using Mahalanobis distance as the Caesars sensitivity index (CSI). 。 7. The method for capturing and analyzing early jamming symptoms of switchgear circuit breakers according to claim 1, characterized in that, The tracking and analysis of the Caesars sensitivity index series using linear regression trend prediction includes: The sticking sensitivity index, calculated based on the effective waveform segment signal of each action, is recorded in order of the number of actions to form a sticking sensitivity index-number of actions trend sequence; Linear regression was used to fit the stickiness sensitivity index of the most recent actions to obtain the recent growth slope. When the recent growth slope β is confirmed to be greater than the set slope threshold twice in a row, and the jamming sensitivity index CSI is between the light alarm threshold and the heavy alarm threshold, a first-level lubrication maintenance recommendation is output. When the sensitivity index (CSI) is greater than the critical alarm threshold, a secondary mechanism maintenance alarm is output.

8. The method for capturing and analyzing early jamming symptoms of switchgear circuit breakers according to claim 1, characterized in that, Multidimensional eigenvector contribution tracing analysis includes: After identifying the anomaly, backtrack the contribution of each dimension of the current anomaly's multidimensional feature vector; If the sum of the deviation contributions of the mean energy of high-frequency vibration and the standard deviation of frequency is the largest, the jamming is determined to originate from the sliding friction pair of the iron core or push rod; if the sum of the deviation contributions of the cumulative change in curvature and the micro deviation rate of duration is the largest, the jamming is determined to originate from the nonlinear jamming of the transmission link or spring mechanism.

9. The method for capturing and analyzing early jamming symptoms of switchgear circuit breakers according to claim 8, characterized in that, The contribution of the i-th feature of a multidimensional feature vector to the Mahalanobis distance. Determined by the product of the deviation values ​​of each dimension's features and the inverse covariance matrix: ； in, Multidimensional feature vectors The i-th and j-th elements, Multidimensional feature vectors The mean vector, Here is the covariance matrix, and CSI is the Cauchy sensitivity index.

10. A device for capturing and analyzing early signs of jamming in switchgear circuit breakers, comprising: At least one processing unit, wherein the processing unit is connected to a storage unit and a data acquisition unit via a bus unit, characterized in that the processing unit implements the method for capturing and analyzing early jamming symptoms of a switchgear circuit breaker as described in any one of claims 1-9 by running a computer-executable program stored in the storage unit.

Images