A method and system for monitoring faults of vacuum AC contactors

Abstract

The present application relates to a vacuum AC contactor fault monitoring method and system. The method comprises: collecting historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, obtaining a sample matrix under each working condition for preprocessing, data enhancement and expansion, and obtaining each expanded sample matrix; extracting the final subspace matrix determined by the subspace basis vector according to all the expanded sample matrices, determining the information useful for state evaluation and the information irrelevant to state evaluation in the expanded sample matrix of the normal working condition; extracting the information useful for the fault diagnosis process from the information irrelevant to the state evaluation and adding it to the calculation of the useful information for state evaluation T 2 Statistics, get T 2 The fault monitoring threshold is determined by statistical samples. The fault monitoring threshold is used to monitor whether the monitored vacuum AC contactor fails, which can improve the accuracy of fault monitoring for vacuum contactors with complex structures and working conditions.

Description

Technical Field

[0001] The present application relates to the field of electric power technology, and in particular to a vacuum AC contactor fault monitoring method, device and system. Background Art

[0002] Vacuum AC contactors are commonly used in industrial fields, such as mining, metallurgy, electricity, and shipbuilding. They use the arc extinction characteristics in a vacuum to achieve rapid disconnection of high-voltage circuits, thereby protecting circuits and equipment. Failure of vacuum AC contactors may result in the inability to properly control the flow of power, affecting the reliable and continuous operation of electrical equipment, and may cause more serious equipment damage or safety accidents due to the inability to disconnect the circuit in an emergency.

[0003] In view of its multiple functions in high-voltage electrical equipment and its core role in the stability of the power system, as well as the high investment cost of related equipment, it is particularly important to conduct research on fault prevention and performance monitoring of vacuum AC contactors. By ensuring its efficient and reliable operation, it can not only improve the operating efficiency of the equipment, but also reduce potential safety risks and protect the safety of personnel and property. Vacuum AC contactors are more commonly used in high-voltage scenarios. Compared with ordinary AC contactors, vacuum AC contactors have stronger arc-breaking capabilities and faster response speeds. In order to ensure the reliability of each interruption, the mechanical structure of the vacuum AC contactor is also more complex, which will lead to a higher probability of failure of the vacuum AC contactor, but there is less research on its fault detection.

[0004] At present, there are methods for monitoring the faults of vacuum AC contactors by using current signals through machine learning methods such as neural networks and random forests. However, the existing fault monitoring methods for vacuum AC contactors still have the following problems: 1. The feature extraction of the contactor current signal is insufficient. Usually only the original current value, the time of closing and opening the switch, etc. are used, resulting in the selected features being insufficient to reflect the fault changes. When applied to vacuum contactors with complex structures and working conditions, it is impossible to obtain high-accuracy fault monitoring results. Summary of the invention

[0005] Based on this, it is necessary to provide a vacuum AC contactor fault monitoring method, device and system that can improve the accuracy of fault monitoring of vacuum contactors with complex structures and working conditions in order to address the above technical problems.

[0006] A vacuum AC contactor fault monitoring method, the method comprising:

[0007] Collect the historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, and obtain the sample matrix X under each working condition c =[x1 c ,x2 c,…,x K c ] T , where X c is the sample matrix under the cth operating condition, x k c is the kth sample data under the cth working condition, k∈1, 2, 3...K, K is the number of samples under the cth working condition, the superscript T represents transposition, and the different working conditions include multiple fault conditions and one normal working condition;

[0008] Intercept the sample matrix X under each working condition c The data of the rising response period in the sample data in the , and taking the minimum length sample data as the benchmark, the sample data under each working condition are processed with equal length to obtain each processed sample matrix X1 c =[x1 c ,x2 c ,…,x M c ] T , where X1 c is the sample matrix X c The processed sample matrix, M is the processed sample matrix X1 c The number of samples, M≤K;

[0009] Calculate each processed sample matrix X1 c The difference, cumulative area, gradient, mean, and variance of the sample data in each processed sample matrix X1 c Perform data enhancement and standardization to obtain each enhanced sample matrix Among them, X2 c is the processed sample matrix X1 c The enhanced sample matrix, x i c is the i-th sample data under the c-th working condition, is the difference of the ith sample data under the cth operating condition, is the cumulative area of ​​the ith sample data under the cth operating condition, is the gradient of the i-th sample data under the c-th operating condition, is the mean of the i-th sample data under the c-th operating condition, is the variance of the i-th sample data under the c-th operating condition, i∈1, 2, 3…M;

[0010] Select an enhanced sample matrix to analyze and determine the dynamic relationship number, and use the dynamic data processing method to process each of the enhanced sample matrices X2 cExpand and obtain each expanded sample matrix;

[0011] Extract subspace basis vectors according to all expanded sample matrices to obtain the final subspace matrix;

[0012] According to the final subspace matrix, the expanded sample matrix of the normal operating state is divided to determine the information useful for state evaluation and the information irrelevant for state evaluation in the expanded sample matrix of the normal operating state;

[0013] After subdividing the information irrelevant to the state evaluation in the expanded sample matrix of the normal working condition according to the final subspace matrix, the information useful for the fault diagnosis process is extracted and added to the useful information for the state evaluation, so as to obtain the supplemented information useful for the state evaluation;

[0014] The supplemented beneficial information for status evaluation is processed by PCA, and the matrix after PCA processing is calculated and T 2 Statistics, get T 2 Statistics sample, according to the T obtained 2 The statistical sample determines the fault monitoring threshold;

[0015] Collect the opening and closing current data of the monitored vacuum AC contactor in real time, and perform T 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2 Whether the statistic is within the range of the fault monitoring threshold value is determined to determine whether the monitored vacuum AC contactor fails.

[0016] In one embodiment, the dynamic relationship number determined by selecting an enhanced sample matrix for analysis is determined by using a dynamic data processing method to process each of the enhanced sample matrices X2 c Expand and obtain each expanded sample matrix, including:

[0017] Select an enhanced sample matrix to perform eigendecomposition to determine the dynamic relationship number and analyze the final hysteresis factor;

[0018] According to the final hysteresis factor, a dynamic data processing method is used to process each enhanced sample matrix X2 c Expand to obtain each expanded sample matrix, the expanded sample matrix X l c for:

[0019]

[0020] Among them, X lc is the expanded sample matrix under the cth operating condition.

[0021] In one embodiment, the step of selecting an enhanced sample matrix to perform eigendecomposition to determine the dynamic relationship number and analyzing the final hysteresis factor includes:

[0022] Calculate the covariance matrix of the selected enhanced sample matrix For the covariance matrix The characteristic decomposition is: Obtain the load matrix P = [p1, p2, ..., p6], where p i is the principal component vector, i∈1, 2, 3…6, the eigenvalue matrix Λ=diag{λ1,λ2,…,λ6}, λ i For p i The corresponding eigenvalues ​​are i∈1, 2, 3, ... 6;

[0023] Extract the cumulative contribution rate η of the principal component vector in the load matrix p ≥85% of the principal component vectors, and the extracted loading matrix is ​​obtained Among them, the cumulative contribution rate η p The expression is:

[0024]

[0025] Among them, d is the number of principal components to be saved in the extracted load matrix;

[0026] According to the number of principal components to be stored in the initial extracted load matrix, the initial static relationship number is determined. The expression of the initial static relationship number r(0) is:

[0027] r(0)=v–d

[0028] Among them, v is the initial number of variables, and d is the number of principal components to be stored in the initial extracted load matrix;

[0029] Let the current hysteresis factor l = 1, expand the sample data in the selected enhanced sample matrix, obtain the corresponding current expanded sample matrix, and calculate the covariance matrix of the current expanded sample matrix for eigendecomposition and extract the cumulative contribution rate η p ≥85% of the principal component vectors, determine the number of principal components to be stored in the load matrix after the current expansion and extraction, and calculate the static relationship number after the current expansion. The expression of the static relationship number r(l) after the current expansion is:

[0030] r(l)=v l –d l

[0031] Among them, v l is the number of variables after current expansion, d l The number of principal components to be saved in the current expanded extracted load matrix;

[0032] According to the current expanded static relationship number and the initial static relationship number, the dynamic relationship number recursion formula is used to determine the current expanded dynamic relationship number. The expression of the dynamic relationship number recursion formula is:

[0033]

[0034] Among them, r new (l) is the number of dynamic relations after the current expansion, and r(l) is the number of static relations after the current expansion. When i=0, r new (0) is r(0);

[0035] Let the current hysteresis factor l = 2, expand the sample data in the last expanded sample matrix, obtain the corresponding current expanded sample matrix, and calculate the covariance matrix of the current expanded sample matrix for eigendecomposition and extract the cumulative contribution rate η p ≥85% of the principal component vectors, determine the number of principal components to be stored in the load matrix after the current expansion and extraction, calculate the static relationship number after the current expansion, and determine the dynamic relationship number after the current expansion by using the dynamic relationship number recursive formula according to the static relationship number after the current expansion and the initial static relationship number;

[0036] When the current hysteresis factor l=2 is judged, whether the current expanded dynamic relationship number is less than 0, if it is less than 0, the final hysteresis factor is determined to be l-1, if it is not less than 0, the current hysteresis factor l=l+1 is continued, and the sample data in the last expanded sample matrix is ​​expanded repeatedly, and the current expanded sample matrix is ​​obtained accordingly, and the covariance matrix of the current expanded sample matrix is ​​calculated to perform eigendecomposition and extract the cumulative contribution rate η p ≥85% of the principal component vectors, determine the number of principal components to be stored in the load matrix after the current expansion and extraction, calculate the static relationship number after the current expansion, and use the dynamic relationship number recursive formula to determine the dynamic relationship number after the current expansion according to the static relationship number after the current expansion and the initial static relationship number, and judge whether the dynamic relationship number after the current expansion is less than 0. If it is less than 0, determine the final hysteresis factor to be l-1. If it is not less than 0, continue to set the current hysteresis factor l=l+1 until the dynamic relationship number after the current expansion is less than 0, and determine the final hysteresis factor.

[0037] In one embodiment, extracting subspace basis vectors according to all expanded sample matrices to obtain a final subspace matrix includes:

[0038] According to the intermediate basis vector space Solving equations The basis vectors of each subset Expressed as Among them, Q is the number of working conditions, is the intermediate basis vector of the subspace, f∈1,2,3…F, F represents the dimension of the matrix, F = 6(l+1);

[0039] Using the intermediate basis vector space Represents the sample matrix X l c : Substitute into the final subspace matrix P g Solve the solution equation to obtain the final subspace matrix P g , the solution for P g The solution equation is:

[0040]

[0041] Among them, the final subspace P g =[p g,1 p g,2 ,…,p g,F ], p g,f is the final subspace basis vector, f∈1,2,3…F,λ g is a scalar.

[0042] In one embodiment, the expression for dividing the expanded sample matrix of the normal operating state is:

[0043]

[0044] Among them, X l q is the expanded sample matrix of normal working condition, To evaluate irrelevant information for the state, To provide useful information for state evaluation, I is the unit matrix.

[0045] In one embodiment, after subdividing the state evaluation irrelevant information in the expanded sample matrix of the normal operating condition according to the final subspace matrix, extracting the information useful for the fault diagnosis process and adding it to the state evaluation useful information, obtaining the supplemented information useful for the state evaluation, includes:

[0046] The information irrelevant to state evaluation in the expanded sample matrix of the normal working condition is projected onto the final subspace matrix, and the amplitudes of the information irrelevant to state evaluation are compared. Subdivided into in, is a subspace with similar amplitude, is a subspace with dissimilar amplitudes;

[0047] The information irrelevant to the state evaluation in the subspace with dissimilar amplitudes is determined as information beneficial to the fault diagnosis process and added to the beneficial information for the state evaluation to obtain the supplemented information beneficial to the state evaluation. The supplement is helpful for evaluating the status for

[0048] In one embodiment, the T 2 Statistics Sample The expression is:

[0049]

[0050] in, is the matrix after PCA processing, P′ is the extracted load matrix obtained by PCA processing after supplementing the beneficial information for state evaluation, and ∑ is the eigenvalue matrix obtained by PCA processing after supplementing the beneficial information for state evaluation.

[0051] A vacuum AC contactor fault monitoring system, the system comprising:

[0052] The sample acquisition module is used to collect the historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, and obtain the sample matrix X under each working condition c =[x1 c ,x2 c ,…,x K c ] T , where X c is the sample matrix under the cth operating condition, x k c is the kth sample data under the cth working condition, k∈1, 2, 3...K, K is the number of samples under the cth working condition, the superscript T represents transposition, and the different working conditions include multiple fault conditions and one normal working condition;

[0053] The sample preprocessing module is used to intercept the sample matrix X under each working condition. cThe data of the rising response period in the sample data in the , and taking the minimum length sample data as the benchmark, the sample data under each working condition are processed with equal length to obtain each processed sample matrix X1 c =[x1 c ,x2 c ,…,x M c ] T , where X1 c is the sample matrix X c The processed sample matrix, M is the processed sample matrix X1 c The number of samples, M≤K;

[0054] The sample enhancement module is used to calculate each processed sample matrix X1 c The difference, cumulative area, gradient, mean, and variance of the sample data in each processed sample matrix X1 c Perform data enhancement and standardization to obtain each enhanced sample matrix Among them, X2 c is the processed sample matrix X1 c The enhanced sample matrix, x i c is the i-th sample data under the c-th working condition, is the difference of the ith sample data under the cth operating condition, is the cumulative area of ​​the ith sample data under the cth operating condition, is the gradient of the i-th sample data under the c-th operating condition, is the mean of the i-th sample data under the c-th operating condition, is the variance of the i-th sample data under the c-th operating condition, i∈1, 2, 3…M;

[0055] The sample expansion module is used to select an enhanced sample matrix for analysis and determine the dynamic relationship number, and use the dynamic data processing method to process each of the enhanced sample matrices X2 c Expand and obtain each expanded sample matrix;

[0056] A subspace extraction module is used to extract subspace basis vectors according to all the expanded sample matrices to obtain the final subspace matrix;

[0057] A sample partitioning module, used to partition the expanded sample matrix of the normal operating state according to the final subspace matrix, and determine the information useful for state evaluation and the information irrelevant for state evaluation in the expanded sample matrix of the normal operating state;

[0058] A sample subdivision module is used to subdivide the information irrelevant to the state evaluation in the expanded sample matrix of the normal working condition according to the final subspace matrix, extract the information useful for the fault diagnosis process and add it into the useful information for the state evaluation, so as to obtain the supplemented information useful for the state evaluation;

[0059] The statistical analysis module is used to perform PCA processing on the supplemented beneficial information for status evaluation, and calculate the matrix after PCA processing for T 2 Statistics, get T 2 Statistics sample, according to the T obtained 2 The statistical sample determines the fault monitoring threshold;

[0060] The monitoring module is used to collect the opening and closing current data of the monitored vacuum AC contactor in real time, and perform T 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2 Whether the statistic is within the range of the fault monitoring threshold value is determined to determine whether the monitored vacuum AC contactor fails.

[0061] A vacuum AC contactor fault monitoring system, the system comprising: a monitored vacuum AC contactor, a Hall current sensor, an intermediate relay, an A / D data processing chip, a single chip microcomputer and a host computer;

[0062] The monitored vacuum AC contactor is connected to the Hall current sensor and the intermediate relay, the Hall current sensor is connected to the A / D data processing chip, the A / D data processing chip is connected to the single-chip microcomputer, the single-chip microcomputer is connected to the intermediate relay and the host computer,

[0063] The Hall current sensor collects the opening and closing current data of the monitored vacuum AC contactor in real time and transmits it to the A / D data processing chip. The A / D data processing chip transmits the collected opening and closing current data to the single-chip microcomputer. The opening and closing current data is transmitted to the host computer through the single-chip microcomputer. The host computer executes the above-mentioned vacuum AC contactor fault monitoring method according to the collected opening and closing current data to determine whether the monitored vacuum AC contactor has a fault. When it is determined that the monitored vacuum AC contactor has sent a fault, the host computer sends a control instruction to the intermediate relay through the single-chip microcomputer, so that the intermediate relay controls the vacuum AC contactor to disconnect the circuit. The host computer also reminds the operator to take timely maintenance or protection measures to prevent the fault from worsening.

[0064] The above vacuum AC contactor fault monitoring method, device and system collect the historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, obtain the sample matrix under each working condition, and intercept the sample matrix X under each working condition. c The data of the rising response period in the sample data in the sample data is taken as the benchmark, and the sample data under each working condition is processed with equal length to obtain each processed sample matrix, and each processed sample matrix X1 is calculated. c The difference, cumulative area, gradient, mean, and variance of the sample data in each processed sample matrix X1 c Perform data enhancement and standardization to obtain each enhanced sample matrix, select an enhanced sample matrix for analysis and determine the dynamic relationship number, and use the dynamic data processing method to process each enhanced sample matrix X2 c Expand to obtain each expanded sample matrix; extract subspace basis vectors according to all the expanded sample matrices to obtain the final subspace matrix; divide the expanded sample matrix of the normal working condition according to the final subspace matrix to determine the information that is beneficial to the state evaluation and the information that is irrelevant to the state evaluation in the expanded sample matrix of the normal working condition; subdivide the information that is irrelevant to the state evaluation in the expanded sample matrix of the normal working condition according to the final subspace matrix, extract the information that is beneficial to the fault diagnosis process and add it to the beneficial information for the state evaluation, and obtain the supplemented beneficial information for the state evaluation; perform PCA processing on the supplemented beneficial information for the state evaluation, and calculate the matrix after PCA processing for T 2 Statistics, get T 2 Statistics sample, according to the obtained T 2 The fault monitoring threshold is determined by statistical samples. The upper and lower control limits of the fault monitoring threshold are closer to the critical value of the actual fault of the vacuum AC contactor. Then, the opening and closing current data of the monitored vacuum AC contactor are collected in real time. T 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2 Whether the statistic is within the range of the fault monitoring threshold value and whether the monitored vacuum AC contactor has a fault is determined, which can improve the accuracy of fault monitoring for vacuum contactors with complex structures and working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 A schematic diagram of the structure of a vacuum AC contactor fault monitoring system in one embodiment;

[0066] Figure 2 A schematic flow chart of a vacuum AC contactor fault monitoring method in one embodiment;

[0067] Figure 3 Draw a signal waveform diagram according to time-opening and closing current data of sample data in one embodiment;

[0068] Figure 4 is a waveform diagram of upper and lower control limit thresholds in one embodiment;

[0069] Figure 5 is the current T in normal working condition in one embodiment. 2 Waveform diagrams of statistics and fault monitoring thresholds;

[0070] Figure 6 is the current T in an abnormal working state in one embodiment. 2 Waveform diagram of statistics and fault detection thresholds. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0072] The present application provides a vacuum AC contactor fault monitoring method, which can be applied to Figure 1 In the vacuum AC contactor fault monitoring system shown in the figure, the monitored vacuum AC contactor is connected to the Hall current sensor and the intermediate relay, the Hall current sensor is connected to the A / D data processing chip, the A / D data processing chip is connected to the single-chip microcomputer, the single-chip microcomputer is connected to the intermediate relay and the host computer, the Hall current sensor collects the opening and closing current data of the monitored vacuum AC contactor in real time and transmits it to the A / D data processing chip, the A / D data processing chip transmits the collected opening and closing current data to the single-chip microcomputer, the opening and closing current data is transmitted to the host computer through the single-chip microcomputer, and the host computer performs T according to the collected opening and closing current data. 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2Whether the statistic is within the range of the fault monitoring threshold, determine whether the monitored vacuum AC contactor has a fault. When it is determined that the monitored vacuum AC contactor has a fault, the host computer sends a control instruction to the intermediate relay through the single-chip microcomputer, so that the intermediate relay controls the vacuum AC contactor to disconnect the circuit. The host computer also reminds the operator to take timely maintenance or protection measures to prevent the fault from worsening. Among them, AD7606 is used as the A / D data processing chip, which has a 16-bit resolution, 8 analog input channels, and a maximum sampling rate of 200kSPS (200,000 samples per second), which can collect enough data in a very short rise response time. Among them, the host computer can be but not limited to various personal computers, laptops, smart phones, tablets, servers, portable wearable devices, etc.

[0073] In one embodiment, Figure 2 As shown, a vacuum AC contactor fault monitoring method is provided, comprising the following steps:

[0074] Step S220, collecting historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, and obtaining a sample matrix X under each working condition c =[x1 c ,x2 c ,…,x K c ] T , where X c is the sample matrix under the cth operating condition, x k c is the kth sample data under the cth operating condition, k∈1, 2, 3...K, K is the number of samples under the cth operating condition, the superscript T represents transposition, and different operating conditions include multiple fault conditions and one normal operating condition.

[0075] Step S240, intercepting the sample matrix X under each working condition c The data of the rising response period in the sample data in the , and taking the minimum length sample data as the benchmark, the sample data under each working condition are processed with equal length to obtain each processed sample matrix X1 c =[x1 c ,x2 c ,…,x M c ] T , where X1 c is the sample matrix X c The processed sample matrix, M is the processed sample matrix X1 c The number of samples, M≤K.

[0076] Among them, the sample matrix X under each working condition is intercepted c The specific format of the data in the rising response period in the sample data is:

[0077] The sample data in the sample matrix under the cth working condition are plotted as a signal waveform according to the time-opening and closing current data, such as Figure 3 As shown, the extreme value in the opening and closing current data is found, and the extreme value is used to separate the sample data, so as to determine the sample data of the rising response period, that is, the sample data between the start time and the time when the first extreme value appears is the sample data of the rising response period.

[0078] Since the length of the data during the rising response period in the sample matrix under each operating condition is inconsistent, the length of the data during the rising response period in the sample matrix under other operating conditions is truncated based on the shortest length of the data during the rising response period, so that the length of the data during the rising response period in the sample matrix under each operating condition is consistent.

[0079] It should be understood that because important data features are basically contained in the front end of the sample (ie, the rising response period), even if such truncation will lose some features, it can keep the sample period unchanged and retain most of the important information.

[0080] Step S260, calculate each processed sample matrix X1 c The difference, cumulative area, gradient, mean, and variance of the sample data in each processed sample matrix X1 c Perform data enhancement and standardization to obtain each enhanced sample matrix Among them, X2 c is the processed sample matrix X1 c The enhanced sample matrix, x i c is the i-th sample data under the c-th working condition, is the difference of the ith sample data under the cth operating condition, is the cumulative area of ​​the ith sample data under the cth operating condition, is the gradient of the i-th sample data under the c-th operating condition, is the mean of the i-th sample data under the c-th operating condition, is the variance of the i-th sample data under the c-th operating condition, i∈1,2,3…M.

[0081] Among them, the difference of the i-th sample data under the c-th working condition is The calculation formula is:

[0082] Among them, the cumulative area of ​​the i-th sample data under the c-th working condition is The calculation formula is:

[0083] Among them, the gradient of the i-th sample data under the c-th working condition is The calculation formula is:

[0084] Among them, the mean of the i-th sample data under the c-th working condition is The calculation formula is:

[0085] Among them, the variance of the i-th sample data under the c-th working condition is The calculation formula is:

[0086] Among them, for each processed sample matrix X1 c After data enhancement, a data-enhanced matrix is ​​obtained, and sample data of the same dimension in the data-enhanced matrix are standardized to obtain an enhanced sample matrix.

[0087] It should be understood that the sample matrix X1 c There is still only one-dimensional current value, which cannot be used to train the fault monitoring threshold later, so the sample matrix X1 needs to be c After enhancement, the original single current value is changed to a sample matrix containing current magnitude, difference, cumulative area, etc. After this processing, it is the data matrix to be trained.

[0088] Step S280, select an enhanced sample matrix to analyze and determine the dynamic relationship number, and use the dynamic data processing method to process each enhanced sample matrix X2 c Expand and obtain each expanded sample matrix.

[0089] Among them, since the sample matrix has a strong time series feature, it cannot be processed by the general static data processing method. It is necessary to use the dynamic data processing method to expand the enhanced sample matrix. The number of times the enhanced sample matrix is ​​augmented is determined by the dynamic relationship number, that is, the final hysteresis factor l. For each enhanced sample matrix X2 c Expand and obtain each expanded sample matrix.

[0090] It should be understood that the sample data has a strong temporal dynamics and cannot be statically classified. In order to solve this problem, the sample matrix needs to be augmented so that it has data at this moment and the previous moment (it may be more than the previous moment, and the number of augmentations needs to be determined based on calculations). The augmented matrix (i.e., the expanded sample matrix) after processing contains temporal features and can be dynamically processed.

[0091] In one embodiment, an enhanced sample matrix is ​​selected for analysis to determine the dynamic relationship number, and a dynamic data processing method is used to process each enhanced sample matrix X2 c Expand and obtain each expanded sample matrix, including:

[0092] Select an enhanced sample matrix to perform eigendecomposition to determine the dynamic relationship number and analyze the final hysteresis factor;

[0093] According to the final hysteresis factor, the dynamic data processing method is used to process each enhanced sample matrix X2 c Expand and obtain each expanded sample matrix, the expanded sample matrix X l c for:

[0094]

[0095] Among them, X l c is the expanded sample matrix under the cth operating condition.

[0096] In one embodiment, the step of selecting an enhanced sample matrix to perform eigendecomposition to determine the dynamic relationship number and analyzing the final hysteresis factor includes:

[0097] Calculate the covariance matrix of the selected enhanced sample matrix Covariance matrix The characteristic decomposition is: Obtain the load matrix P = [p1, p2, ..., p6], where p i is the principal component vector, i∈1, 2, 3…6, the eigenvalue matrix Λ=diag{λ1,λ2,…,λ6}, λ i For p i The corresponding eigenvalues ​​are i∈1, 2, 3, ... 6;

[0098] Extract the cumulative contribution rate η of the principal component vector in the load matrix p ≥85% of the principal component vectors, and the extracted loading matrix is ​​obtained Among them, the cumulative contribution rate η p The expression is:

[0099]

[0100] Among them, d is the number of principal components to be saved in the extracted load matrix;

[0101] According to the number of principal components to be saved in the initial extracted load matrix, the initial static relationship number is determined. The expression of the initial static relationship number r(0) is:

[0102] r(0)=v–d

[0103] Among them, v is the initial number of variables, and d is the number of principal components to be stored in the initial extracted load matrix;

[0104] Let the current hysteresis factor l = 1, expand the sample data in the selected enhanced sample matrix, obtain the corresponding current expanded sample matrix, and calculate the covariance matrix of the current expanded sample matrix for eigendecomposition and extract the cumulative contribution rate η p ≥85% of the principal component vectors, determine the number of principal components to be saved in the load matrix after the current expansion and extraction, calculate the static relationship number after the current expansion, and the expression of the static relationship number r(l) after the current expansion is:

[0105] r(l)=v l –d l

[0106] Among them, v l is the number of variables after current expansion, d l The number of principal components to be saved in the current expanded extracted load matrix;

[0107] According to the current expanded static relationship number and the initial static relationship number, the dynamic relationship number recursion formula is used to determine the current expanded dynamic relationship number. The expression of the dynamic relationship number recursion formula is:

[0108]

[0109] Among them, r new (l) is the number of dynamic relations after the current expansion, and r(l) is the number of static relations after the current expansion. When i=0, r new (0) is r(0);

[0110] Let the current hysteresis factor l = 2, expand the sample data in the last expanded sample matrix, obtain the corresponding current expanded sample matrix, and calculate the covariance matrix of the current expanded sample matrix for eigendecomposition and extract the cumulative contribution rate η p≥85% of the principal component vectors, determine the number of principal components to be saved in the load matrix after the current expansion, calculate the static relationship number after the current expansion, and use the dynamic relationship number recursive formula to determine the dynamic relationship number after the current expansion based on the static relationship number after the current expansion and the initial static relationship number;

[0111] When the current hysteresis factor l=2 is judged, whether the current expanded dynamic relationship number is less than 0, if it is less than 0, the final hysteresis factor is determined to be l-1, if it is not less than 0, the current hysteresis factor l=l+1 is continued, and the sample data in the last expanded sample matrix is ​​repeatedly expanded to obtain the current expanded sample matrix, and the covariance matrix of the current expanded sample matrix is ​​calculated for eigendecomposition and the cumulative contribution rate η is extracted p ≥85% of the principal component vectors, determine the number of principal components to be saved in the load matrix after the current expansion and extraction, calculate the static relationship number after the current expansion, and use the dynamic relationship number recursive formula to determine the dynamic relationship number after the current expansion according to the static relationship number after the current expansion and the initial static relationship number. Determine whether the dynamic relationship number after the current expansion is less than 0. If it is less than 0, determine the final hysteresis factor to be l-1. If it is not less than 0, continue to set the current hysteresis factor l=l+1 until the dynamic relationship number after the current expansion is less than 0, and determine the final hysteresis factor.

[0112] Step S300: extract subspace basis vectors according to all expanded sample matrices to obtain a final subspace matrix.

[0113] Among them, since the collected opening and closing current data is measured based on the Hall current sensor and has a fault data set with different working conditions, it is not possible to accurately judge the fault situation by simply relying on the current size for online monitoring, so the original data is enhanced, but the enhanced data is a multi-sample and multi-dimensional data set, but it is difficult to determine whether the features in each dimension can play a role in judging whether a fault occurs during operation, so it is necessary to remove elements that are irrelevant to the fault diagnosis process. These elements are often common variable correlations between different sets. In order to determine the subspace of common variable correlations between different sets, the subspace basis vectors can be extracted according to all the expanded sample matrices to obtain the final subspace matrix, and then determine the useful information for state evaluation.

[0114] In one embodiment, subspace basis vectors are extracted according to all expanded sample matrices to obtain a final subspace matrix, including:

[0115] According to the intermediate basis vector space Solving equations The basis vectors of each subset Expressed as Among them, Q is the number of working conditions, is the intermediate basis vector of the subspace, f∈1,2,3…F, F represents the dimension of the matrix, F = 6(l+1);

[0116] Using the intermediate basis vector space Represents the sample matrix X l c : Substitute into the final subspace matrix P g Solve the solution equation to obtain the final subspace matrix P g , solve for P g The solution equation is:

[0117]

[0118] Among them, the final subspace P g =[p g,1 p g,2 ,…,p g,F ], p g,f is the final subspace basis vector, f∈1,2,3…F,λ g is a scalar.

[0119] Step S320 , dividing the expanded sample matrix of the normal operating state according to the final subspace matrix, and determining the information useful for state evaluation and the information irrelevant to state evaluation in the expanded sample matrix of the normal operating state.

[0120] In one embodiment, the expression for partitioning the expanded sample matrix of the normal operating state is:

[0121]

[0122] Among them, X l q is the expanded sample matrix of normal working condition, To evaluate irrelevant information for the state, To provide useful information for state evaluation, I is the unit matrix.

[0123] Step S340, after subdividing the information irrelevant to the state evaluation in the expanded sample matrix of the normal operating condition according to the final subspace matrix, extracting the information useful for the fault diagnosis process and adding it into the information useful for the state evaluation, thereby obtaining the supplemented information useful for the state evaluation.

[0124] In one embodiment, after the state evaluation irrelevant information is subdivided in the expanded sample matrix of the normal working condition according to the final subspace matrix, the information useful for the fault diagnosis process is extracted and added to the state evaluation useful information, and the supplemented useful information for the state evaluation is obtained, including:

[0125] Project the information irrelevant to state evaluation in the expanded sample matrix of the normal working condition to the final subspace matrix, compare the amplitudes of the information irrelevant to state evaluation, and Subdivided into in, is a subspace with similar amplitude, is a subspace with dissimilar amplitudes;

[0126] The information irrelevant to the state evaluation in the subspace with dissimilar amplitude is determined as information beneficial to the fault diagnosis process and added to the beneficial information of the state evaluation to obtain the supplemented information beneficial to the state evaluation. Supplemented with useful information for status evaluation for

[0127] Among them, if the amplitudes are similar (that is, the amplitude difference is less than or equal to the threshold value, which is irrelevant information for state evaluation), it can be considered that the variation information in this direction is the common variation information of each operating condition, and cannot play a role in distinguishing normal operating conditions or abnormal operating conditions. On the contrary, if the amplitude difference is large (that is, the amplitude difference is greater than the threshold value, which is irrelevant information for state evaluation), it can be considered as a part that is beneficial to the fault diagnosis process and state evaluation can be performed.

[0128] Step S360, PCA processing is performed on the supplemented state evaluation useful information, and the matrix after PCA processing is calculated and T 2 Statistics, get T 2 Statistics sample, according to the obtained T 2 The statistical sample determines the fault detection threshold.

[0129] In one embodiment, T 2 Statistics Sample The expression is:

[0130]

[0131] in, is the matrix after PCA processing, P′ is the extracted load matrix obtained by PCA processing after supplementing the beneficial information for state evaluation, and ∑ is the eigenvalue matrix obtained by PCA processing after supplementing the beneficial information for state evaluation.

[0132] Among them, T 2 Statistics Sample is a vector with time series characteristics, and a sliding window is used to process T 2 Statistics Sample The data that can distinguish the top 10% from the bottom 10% in each window are selected as the upper and lower control limit thresholds of the window. Each window corresponds to a monitoring stage, such as Figure 4 The upper and lower control limit threshold waveforms are shown.

[0133] Step S380, real-time acquisition of the switching current data of the monitored vacuum AC contactor, and T 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2 Whether the statistic is within the range of the fault monitoring threshold value determines whether the monitored vacuum AC contactor fails.

[0134] It should be understood that, in the monitoring stage, the current T 2 The process of obtaining statistics is similar to that of obtaining T 2 The statistical samples are processed in the same way, only the data are different, which will not be elaborated here.

[0135] According to the monitoring stage in which the opening and closing current data of the monitored vacuum AC contactor are collected, the fault monitoring threshold of the monitoring stage is determined, so as to 2 Whether the statistic is within the range of the fault monitoring threshold value of the monitoring stage determines whether the vacuum AC contactor currently being monitored has a fault.

[0136] In one example, if Figure 5 The current T under normal operating conditions is shown 2 The waveform diagram of the statistics and fault monitoring threshold shows that the current T 2 If the waveform of the statistic can be well matched with the set upper and lower control limit threshold waveforms, it can be judged that it is in normal working condition; Figure 6 The current T under abnormal operating conditions shown 2 The waveform diagram of the statistics and fault monitoring threshold shows that the current T 2 The waveform of the statistic shows a large deviation from the range of the fault monitoring threshold value in the monitoring stage, which indicates that a fault has occurred.

[0137] The above vacuum AC contactor fault monitoring method collects the historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, obtains the sample matrix under each working condition, and intercepts the sample matrix X under each working condition. cThe data of the rising response period in the sample data in the sample data is taken as the benchmark, and the sample data under each working condition is processed with equal length to obtain each processed sample matrix, and each processed sample matrix X1 is calculated. c The difference, cumulative area, gradient, mean, and variance of the sample data in each processed sample matrix X1 c Perform data enhancement and standardization to obtain each enhanced sample matrix, select an enhanced sample matrix for analysis and determine the dynamic relationship number, and use the dynamic data processing method to process each enhanced sample matrix X2 c Expand to obtain each expanded sample matrix; extract subspace basis vectors according to all the expanded sample matrices to obtain the final subspace matrix; divide the expanded sample matrix of the normal working condition according to the final subspace matrix to determine the information that is beneficial to the state evaluation and the information that is irrelevant to the state evaluation in the expanded sample matrix of the normal working condition; subdivide the information that is irrelevant to the state evaluation in the expanded sample matrix of the normal working condition according to the final subspace matrix, extract the information that is beneficial to the fault diagnosis process and add it to the beneficial information for the state evaluation, and obtain the supplemented beneficial information for the state evaluation; perform PCA processing on the supplemented beneficial information for the state evaluation, and calculate the matrix after PCA processing for T 2 Statistics, get T 2 Statistics sample, according to the obtained T 2 The fault monitoring threshold is determined by statistical samples. The upper and lower control limits of the fault monitoring threshold are closer to the critical value of the actual fault of the vacuum AC contactor. Then, the opening and closing current data of the monitored vacuum AC contactor are collected in real time. T 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2 Whether the statistic is within the range of the fault monitoring threshold value and whether the monitored vacuum AC contactor has a fault is determined, which can improve the accuracy of fault monitoring for vacuum contactors with complex structures and working conditions.

[0138] It should be understood that although Figure 2 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 2At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0139] In one embodiment, a vacuum AC contactor fault monitoring system includes:

[0140] The sample acquisition module is used to collect the historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, and obtain the sample matrix X under each working condition c =[x1 c ,x2 c ,…,x K c ] T , where X c is the sample matrix under the cth operating condition, x k c is the kth sample data under the cth working condition, k∈1, 2, 3…K, K is the number of samples under the cth working condition, the superscript T represents transposition, and different working conditions include multiple fault conditions and one normal working condition;

[0141] Sample preprocessing module, used to intercept the sample matrix X under each working condition c The data of the rising response period in the sample data in the , and taking the minimum length sample data as the benchmark, the sample data under each working condition are processed with equal length to obtain each processed sample matrix X1 c =[x1 c ,x2 c ,…,x M c ] T , where X1 c is the sample matrix X c The processed sample matrix, M is the processed sample matrix X1 c The number of samples, M≤K;

[0142] Sample enhancement module, used to calculate each processed sample matrix X1 c The difference, cumulative area, gradient, mean, and variance of the sample data in each processed sample matrix X1 c Perform data enhancement and standardization to obtain each enhanced sample matrix Among them, X2 c is the processed sample matrix X1 cThe enhanced sample matrix, x i c is the i-th sample data under the c-th working condition, is the difference of the ith sample data under the cth operating condition, is the cumulative area of ​​the ith sample data under the cth operating condition, is the gradient of the i-th sample data under the c-th operating condition, is the mean of the i-th sample data under the c-th operating condition, is the variance of the i-th sample data under the c-th operating condition, i∈1, 2, 3…M;

[0143] The sample expansion module is used to select an enhanced sample matrix for analysis and determine the dynamic relationship number, and use the dynamic data processing method to process each enhanced sample matrix X2 c Expand and obtain each expanded sample matrix;

[0144] A subspace extraction module is used to extract subspace basis vectors according to all the expanded sample matrices to obtain the final subspace matrix;

[0145] A sample partitioning module is used to partition the expanded sample matrix of the normal working condition according to the final subspace matrix, and determine the information useful for state evaluation and the information irrelevant for state evaluation in the expanded sample matrix of the normal working condition;

[0146] The sample segmentation module is used to segment the information irrelevant to the state evaluation in the expanded sample matrix of the normal working condition according to the final subspace matrix, extract the information useful for the fault diagnosis process and add it into the useful information for the state evaluation, so as to obtain the supplemented information useful for the state evaluation;

[0147] The statistical analysis module is used to perform PCA processing on the beneficial information of the status evaluation after supplementation, and calculate the matrix after PCA processing for T 2 Statistics, get T 2 Statistics sample, according to the obtained T 2 The statistical sample determines the fault monitoring threshold;

[0148] The monitoring module is used to collect the opening and closing current data of the monitored vacuum AC contactor in real time, and perform T 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2 Whether the statistic is within the range of the fault monitoring threshold value determines whether the monitored vacuum AC contactor fails.

[0149] The specific definition of the vacuum AC contactor fault monitoring system can be found in the definition of the vacuum AC contactor fault monitoring method above, which will not be repeated here. Each module in the above vacuum AC contactor fault monitoring system can be implemented in whole or in part by software, hardware and a combination thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0150] like Figure 1 As shown, in one embodiment, a vacuum AC contactor fault monitoring system includes: a monitored vacuum AC contactor, a Hall current sensor, an intermediate relay, an A / D data processing chip, a single chip microcomputer and a host computer;

[0151] The monitored vacuum AC contactor is connected to the Hall current sensor and the intermediate relay, the Hall current sensor is connected to the A / D data processing chip, the A / D data processing chip is connected to the single-chip microcomputer, the single-chip microcomputer is connected to the intermediate relay and the host computer,

[0152] The Hall current sensor collects the opening and closing current data of the monitored vacuum AC contactor in real time and transmits it to the A / D data processing chip. The A / D data processing chip transmits the collected opening and closing current data to the single-chip microcomputer, and the opening and closing current data is transmitted to the host computer through the single-chip microcomputer. The host computer executes the above-mentioned vacuum AC contactor fault monitoring method according to the collected opening and closing current data to determine whether the monitored vacuum AC contactor has a fault. When it is determined that the monitored vacuum AC contactor has sent a fault, the host computer sends a control instruction to the intermediate relay through the single-chip microcomputer, so that the intermediate relay controls the vacuum AC contactor to disconnect the circuit. The host computer also reminds the operator to take timely maintenance or protection measures to prevent the fault from worsening.

[0153] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0154] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A vacuum AC contactor fault monitoring method, It is characterized in that The method comprises: Collect the historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, and obtain the sample matrix X under each working condition c =[x 1 c , x 2 c , …, x K c ] T , where X c is the sample matrix under the cth operating condition, x k c is the kth sample data under the cth working condition, k∈1, 2, 3...K, K is the number of samples under the cth working condition, the superscript T represents transposition, and the different working conditions include multiple fault working conditions and one normal working condition; Intercept the sample matrix X under each working condition c The sample data in the rising response period are taken as the benchmark, and the sample data under each working condition are processed with equal length to obtain each processed sample matrix X 1 c =[x 1 c , x 2 c , …, x M c ] T , where X 1 c is the sample matrix X c The processed sample matrix, M is the processed sample matrix X 1 c The number of samples, M≤K; Calculate each of the processed sample matrix X 1 c The difference, cumulative area, gradient, mean, and variance of the sample data in each processed sample matrix X 1 c Perform data enhancement and standardization to obtain each enhanced sample matrix Among them, X 2 c is the processed sample matrix X 1 c The enhanced sample matrix, x i c is the i-th sample data under the c-th working condition, is the difference of the ith sample data under the cth operating condition, is the cumulative area of ​​the ith sample data under the cth operating condition, is the gradient of the i-th sample data under the c-th operating condition, is the mean of the i-th sample data under the c-th operating condition, is the variance of the i-th sample data under the c-th operating condition, i∈1,2,3...M; Select an enhanced sample matrix to analyze and determine the dynamic relationship number, and use the dynamic data processing method to process each of the enhanced sample matrices X 2 c Expand and obtain each expanded sample matrix; Extract subspace basis vectors according to all expanded sample matrices to obtain the final subspace matrix; According to the final subspace matrix, the expanded sample matrix of the normal operating state is divided to determine the information useful for state evaluation and the information irrelevant for state evaluation in the expanded sample matrix of the normal operating state; After subdividing the information irrelevant to the state evaluation in the expanded sample matrix of the normal working condition according to the final subspace matrix, the information useful for the fault diagnosis process is extracted and added to the useful information for the state evaluation, so as to obtain the supplemented information useful for the state evaluation; The supplemented beneficial information for status evaluation is processed by PCA, and the matrix after PCA processing is calculated and T 2 Statistics, get T 2 Statistics sample, according to the T obtained 2 The statistical sample determines the fault monitoring threshold; Collect the opening and closing current data of the monitored vacuum AC contactor in real time, and perform T 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2 Whether the statistic is within the range of the fault monitoring threshold value is determined to determine whether the monitored vacuum AC contactor fails.

2. The method according to claim 1, It is characterized in that The dynamic relationship number is determined by selecting an enhanced sample matrix for analysis, and the dynamic data processing method is used to process each of the enhanced sample matrices X 2 c Expand and obtain each expanded sample matrix, including: Select an enhanced sample matrix to perform eigendecomposition to determine the dynamic relationship number and analyze the final hysteresis factor; According to the final hysteresis factor, a dynamic data processing method is used to process each enhanced sample matrix X 2 c Expand to obtain each expanded sample matrix, the expanded sample matrix X l c for: Among them, X l c is the expanded sample matrix under the cth operating condition, and l is the current hysteresis factor.

3. The method according to claim 2, It is characterized in that The step of selecting an enhanced sample matrix to perform eigendecomposition to determine the dynamic relationship number and analyzing the final hysteresis factor includes: Calculate the covariance matrix of the selected enhanced sample matrix For the covariance matrix The characteristic decomposition is: Obtain the load matrix P = [p 1 , p 2 ,…,p 6 ], where p i is the principal component vector, i∈1, 2, 3...6, the eigenvalue matrix Λ=diag{λ 1 ,λ 2 , ..., λ 6 },λ i For p i The corresponding eigenvalues ​​are i∈1, 2, 3, ... 6; Extract the cumulative contribution rate η of the principal component vector in the load matrix p ≥85% of the principal component vectors, and the extracted loading matrix is ​​obtained Among them, the cumulative contribution rate η p The expression is: Among them, d is the number of principal components to be saved in the extracted load matrix; According to the number of principal components to be stored in the initial extracted load matrix, the initial static relationship number is determined. The expression of the initial static relationship number r(0) is: r(0)=vd Among them, v is the initial number of variables, and d is the number of principal components to be stored in the initial extracted load matrix; Let the current hysteresis factor l = 1, expand the sample data in the selected enhanced sample matrix, obtain the corresponding current expanded sample matrix, and calculate the covariance matrix of the current expanded sample matrix for eigendecomposition and extract the cumulative contribution rate η p ≥85% of the principal component vectors, determine the number of principal components to be stored in the load matrix after the current expansion and extraction, and calculate the static relationship number after the current expansion. The expression of the static relationship number r(l) after the current expansion is: r(l)=v l -d l Among them, v l is the number of variables after current expansion, d l The number of principal components to be saved in the currently expanded extracted load matrix; According to the current expanded static relationship number and the initial static relationship number, the dynamic relationship number recursion formula is used to determine the current expanded dynamic relationship number. The expression of the dynamic relationship number recursion formula is: Among them, r new (l) is the number of dynamic relations after the current expansion, and r(l) is the number of static relations after the current expansion. When i=0, r new (0) is r(0); Let the current hysteresis factor l = 2, expand the sample data in the last expanded sample matrix, obtain the corresponding current expanded sample matrix, and calculate the covariance matrix of the current expanded sample matrix for eigendecomposition and extract the cumulative contribution rate η p ≥85% of the principal component vectors, determine the number of principal components to be stored in the load matrix after the current expansion and extraction, calculate the static relationship number after the current expansion, and determine the dynamic relationship number after the current expansion by using the dynamic relationship number recursive formula according to the static relationship number after the current expansion and the initial static relationship number; When the current hysteresis factor l=2 is judged, whether the current expanded dynamic relationship number is less than 0, if it is less than 0, the final hysteresis factor is determined to be l-1, if it is not less than 0, the current hysteresis factor l=l+1 is continued, and the sample data in the last expanded sample matrix is ​​expanded repeatedly, and the current expanded sample matrix is ​​obtained accordingly, and the covariance matrix of the current expanded sample matrix is ​​calculated to perform eigendecomposition and extract the cumulative contribution rate η p ≥85% of the principal component vectors, determine the number of principal components to be stored in the load matrix after the current expansion and extraction, calculate the static relationship number after the current expansion, and use the dynamic relationship number recursive formula to determine the dynamic relationship number after the current expansion according to the static relationship number after the current expansion and the initial static relationship number, and judge whether the dynamic relationship number after the current expansion is less than 0. If it is less than 0, determine the final hysteresis factor to be l-1. If it is not less than 0, continue to set the current hysteresis factor l=l+1 until the dynamic relationship number after the current expansion is less than 0, and determine the final hysteresis factor.

4. The method according to claim 3, It is characterized in that The method of extracting subspace basis vectors according to all the expanded sample matrices to obtain the final subspace matrix includes: According to the intermediate basis vector space Solving equations The basis vectors of each subset Expressed as Among them, Q is the number of working conditions, is the intermediate basis vector of the subspace, f∈1,2,3...F, F represents the dimension of the matrix, F = 6(l+1); Using the intermediate basis vector space Represents the sample matrix X l c : Substitute into the final subspace matrix P g Solve the solution equation to obtain the final subspace matrix P g , the solution for P g The solution equation is: Among them, the final subspace P g =[p g，1 p g，2 ,…,p g，F ], p g，f is the final subspace basis vector, f∈1,2,3...F,λ g is a scalar.

5. The method according to claim 4, It is characterized in that The expression for dividing the expanded sample matrix of the normal working condition is: Among them, X l q is the expanded sample matrix of normal working condition, To evaluate irrelevant information for the state, To provide useful information for state evaluation, I is the unit matrix.

6. The method according to claim 5, It is characterized in that After subdividing the state evaluation irrelevant information in the sample matrix expanded according to the final subspace matrix of the normal working condition, extracting the information useful for the fault diagnosis process and adding it into the state evaluation useful information, the supplemented information useful for the state evaluation is obtained, including: The information irrelevant to the state evaluation in the expanded sample matrix of the normal working condition is projected onto the final subspace matrix, and the amplitudes of the information irrelevant to the state evaluation are compared. Subdivided into in, is a subspace with similar amplitude, is a subspace with dissimilar amplitudes; The information irrelevant to the state evaluation in the subspace with dissimilar amplitudes is determined as information beneficial to the fault diagnosis process and added to the beneficial information for the state evaluation to obtain the supplemented information beneficial to the state evaluation. The supplement is helpful for evaluating the status for 7. The method according to claim 6, It is characterized in that The T2 statistic sample The expression is: in, is the matrix after PCA processing, P′ is the extracted load matrix obtained by PCA processing after supplementing the beneficial information for state evaluation, and ∑ is the eigenvalue matrix obtained by PCA processing after supplementing the beneficial information for state evaluation.

8. A vacuum AC contactor fault monitoring device, It is characterized in that The device comprises: The sample acquisition module is used to collect the historical opening and closing current data of the vacuum AC contactor under different working conditions as sample data, and obtain the sample matrix X under each working condition c =[x 1 c , x 2 c , …, x K c ] T , where X c is the sample matrix under the cth operating condition, x k c is the kth sample data under the cth working condition, k∈1, 2, 3...K, K is the number of samples under the cth working condition, the superscript T represents transposition, and the different working conditions include multiple fault working conditions and one normal working condition; The sample preprocessing module is used to intercept the sample matrix X under each working condition. c The sample data in the rising response period are taken as the benchmark, and the sample data under each working condition are processed with equal length to obtain each processed sample matrix X 1 c =[x 1 c , x 2 c , …, x M c ] T , where X 1 c is the sample matrix X c The processed sample matrix, M is the processed sample matrix X 1 c The number of samples, M≤K; The sample enhancement module is used to calculate each processed sample matrix X 1 c The difference, cumulative area, gradient, mean, and variance of the sample data in each processed sample matrix X 1 c Perform data enhancement and standardization to obtain each enhanced sample matrix Among them, X 2 c is the processed sample matrix X 1 c The enhanced sample matrix, x i c is the i-th sample data under the c-th working condition, is the difference of the ith sample data under the cth operating condition, is the cumulative area of ​​the ith sample data under the cth operating condition, is the gradient of the i-th sample data under the c-th operating condition, is the mean of the i-th sample data under the c-th operating condition, is the variance of the i-th sample data under the c-th operating condition, i∈1,2,3...M; The sample expansion module is used to select an enhanced sample matrix for analysis and determine the dynamic relationship number, and use the dynamic data processing method to process each of the enhanced sample matrices X 2 c Expand and obtain each expanded sample matrix; A subspace extraction module is used to extract subspace basis vectors according to all the expanded sample matrices to obtain the final subspace matrix; A sample partitioning module, used to partition the expanded sample matrix of the normal operating state according to the final subspace matrix, and determine the information useful for state evaluation and the information irrelevant for state evaluation in the expanded sample matrix of the normal operating state; A sample subdivision module is used to subdivide the information irrelevant to the state evaluation in the expanded sample matrix of the normal working condition according to the final subspace matrix, extract the information useful for the fault diagnosis process and add it into the useful information for the state evaluation, so as to obtain the supplemented information useful for the state evaluation; The statistical analysis module is used to perform PCA processing on the supplemented beneficial information for status evaluation, calculate the matrix after PCA processing, and perform T 2 Statistics, get T 2 Statistics sample, according to the T obtained 2 The statistical sample determines the fault monitoring threshold; The monitoring module is used to collect the opening and closing current data of the monitored vacuum AC contactor in real time, and perform T 2 Statistics calculation, get the current T 2 Statistics, based on the current T 2 Whether the statistic is within the range of the fault monitoring threshold value is determined to determine whether the monitored vacuum AC contactor fails.

9. A vacuum AC contactor fault monitoring system, It is characterized in that The system comprises: a monitored vacuum AC contactor, a Hall current sensor, an intermediate relay, an A / D data processing chip, a single chip microcomputer and a host computer; The monitored vacuum AC contactor is connected to the Hall current sensor and the intermediate relay, the Hall current sensor is connected to the A / D data processing chip, the A / D data processing chip is connected to the single-chip microcomputer, the single-chip microcomputer is connected to the intermediate relay and the host computer, The Hall current sensor collects the opening and closing current data of the monitored vacuum AC contactor in real time and transmits it to the A / D data processing chip. The A / D data processing chip transmits the collected opening and closing current data to the single-chip microcomputer. The opening and closing current data is transmitted to the host computer through the single-chip microcomputer. The host computer executes the vacuum AC contactor fault monitoring method according to any one of claims 1 to 7 according to the collected opening and closing current data to determine whether the monitored vacuum AC contactor has a fault. When it is determined that the monitored vacuum AC contactor has sent a fault, the host computer sends a control instruction to the intermediate relay through the single-chip microcomputer, so that the intermediate relay controls the vacuum AC contactor to disconnect the circuit. The host computer also reminds the operator to take timely maintenance or protection measures to prevent the fault from worsening.

Images