A method and system for detecting circuit faults of low voltage circuit breaker
The output current of the low-voltage circuit breaker circuit is denoised through the SVD decomposition algorithm, which solves the problem that noise affects current monitoring accuracy and improves the denoising accuracy and monitoring reliability of current data.
Patent Information
- Application Number
- CN202411273901.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-12
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-09-12
AI Technical Summary
Noise data exists in the output current of the low-voltage circuit breaker circuit, which affects the accuracy of current safety monitoring, resulting in possible misjudgment and equipment damage or accidents.
The SVD decomposition algorithm is used to denoise the current data output from the low-voltage circuit breaker circuit. By fitting the current sinusoidal curve, correcting the initial singular value matrix, and updating the intermediate singular value matrix, the difference between noise and normal data is highlighted to achieve denoise.
It improves the accuracy of current data denoising, reduces noise misjudgment, and ensures the reliability and safety of current monitoring.
Smart Images

Figure CN118777860B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of electrical variable measurement, and in particular to a method and system for detecting circuit faults of a low-voltage circuit breaker. Background Art
[0002] Low-voltage circuit breakers are an important part of the power system. The normal and stable operation of low-voltage circuit breakers plays an irreplaceable protective role in ensuring the stable and safe operation of the power system, such as overload, short circuit, leakage protection, etc. Therefore, it is necessary to effectively detect the low-voltage circuit breaker circuit to reduce the operation risk of the low-voltage circuit breaker. The detection objects of the low-voltage circuit breaker circuit include electrical variables such as voltage, current, and resistance.
[0003] Low-voltage circuit breakers can not only connect and segment normal load current and overload current, but also connect and distinguish short-circuit current switch devices. Therefore, by detecting the output current of the low-voltage circuit breaker circuit, the operating status of various devices on the circuit can be monitored in real time, so as to obtain monitoring information including load conditions, operating temperature, current waveform, etc. When the output current of the low-voltage circuit breaker circuit is too large, it is necessary to output an alarm signal in time and take timely protection measures to prevent equipment damage or accidents.
[0004] The general method of detecting the output current is to use devices such as current sensors or Hall sensors to directly monitor the state of the output current of the low-voltage circuit breaker circuit. Since the actual operation process of the low-voltage circuit breaker is not completely ideal, there will be noise data in the detected output current signal. The noise data will make the waveform of the current signal irregular and unstable, thereby affecting the safety monitoring of the current, which may cause misjudgment of the current monitoring and affect the accuracy of the current safety monitoring. Therefore, denoising the output current of the low-voltage circuit breaker circuit is an urgent problem to be solved. Summary of the invention
[0005] In view of this, in order to solve the technical problem that the accuracy of current safety monitoring is affected by the presence of noise data in the output current of a low-voltage circuit breaker circuit, the present application provides a low-voltage circuit breaker circuit fault detection method and system.
[0006] The technical solutions adopted are as follows:
[0007] In a first aspect, an embodiment of the present application provides a low voltage circuit breaker circuit fault detection method, comprising:
[0008] Acquire output current data of a low voltage circuit breaker circuit, and fit it into a current sinusoidal curve, wherein the current sinusoidal curve includes current sinusoidal periodic curves of multiple time periods;
[0009] According to the difference between the discrete current data outside the current sinusoidal periodic curve and the current sinusoidal periodic curve, the initial singular value matrix of the current data matrix after SVD decomposition is corrected to obtain an intermediate singular value matrix; the current data matrix is converted from the output current data of the low-voltage circuit breaker circuit in the time period corresponding to the current sinusoidal periodic curve;
[0010] The data fluctuation characteristics of the intermediate singular value matrix of the current sinusoidal periodic curve are integrated, and the differences between the intermediate singular value matrices corresponding to several adjacent current sinusoidal periodic curves are updated to obtain the target singular value matrix;
[0011] According to the target singular value matrix, the output current data of the low-voltage circuit breaker circuit is denoised.
[0012] A low-voltage circuit breaker circuit fault detection method provided in an embodiment of the present application has at least the following beneficial effects: denoising the output current of the low-voltage circuit breaker circuit according to the SVD decomposition algorithm. The SVD decomposition algorithm is a non-parameterized method that does not require prior estimation of the noise level or other parameters of the low-voltage circuit breaker circuit output current data. Instead, it adaptively removes noise through the structure of the low-voltage circuit breaker circuit output current data itself, thereby improving denoising flexibility. Moreover, the greater the difference between the noise data point and the current sinusoidal curve, the greater the possibility that it belongs to the noise current data. Therefore, the initial singular value of the current data matrix after SVD decomposition is corrected according to the difference between the discrete current data outside the current sinusoidal periodic curve and the current sinusoidal periodic curve, which can further highlight the difference between the singular values of the noise current data and the normal current data, and prepare for the subsequent singular value correction and update; then, the fluctuation characteristics of the intermediate singular value of the current sinusoidal periodic curve and the difference between the corresponding intermediate singular values of several adjacent current sinusoidal periodic curves are integrated to update the intermediate singular value matrix to obtain the target singular value matrix, so that the target singular value matrix can better highlight the difference between the noise current data and the normal current data, and can improve the accuracy of current data denoising when the output current data of the low-voltage circuit breaker circuit is denoised according to the target singular value, and greatly reduce the misjudgment of the noise current data.
[0013] In combination with the first aspect, in some possible implementations, the process of obtaining the difference between discrete current data outside the current sinusoidal periodic curve and the current sinusoidal periodic curve includes:
[0014] The difference between the candidate discrete current data and the fitting current data at the corresponding position in the current sinusoidal periodic curve is taken as the difference between the candidate discrete current data and the current sinusoidal periodic curve, wherein the candidate discrete current data is any discrete current data.
[0015] In combination with the first aspect above, in some possible implementations, the process of obtaining the intermediate singular value matrix includes:
[0016] Obtaining difference characteristic values between all discrete current data corresponding to the current sinusoidal periodic curve and the current sinusoidal periodic curve, wherein the difference characteristic values are used to characterize the overall difference between all discrete current data and the current sinusoidal periodic curve;
[0017] Performing negative correlation normalization on the difference characteristic value to obtain a correction coefficient;
[0018] The correction coefficients are used as weights to correct the initial singular value matrix to obtain an intermediate singular value matrix.
[0019] The initial singular value matrix is corrected by combining the difference between the discrete current data and the current sinusoidal periodic curve. The initial singular value matrix can be corrected according to the actual situation to improve the correction accuracy.
[0020] In combination with the first aspect above, in some possible implementations, the process of acquiring the current data matrix includes:
[0021] The area of the local output current data in the two-dimensional coordinate system is grid-divided, and the data in each grid is used as an element of the current data matrix, wherein the local output current data is used as the data in the corresponding grid, and the data in other grids are assigned preset values; the local output current data is the output current data of the low-voltage circuit breaker circuit in the corresponding time period of the current sinusoidal periodic curve.
[0022] In combination with the first aspect, in some possible implementations, the process of obtaining the difference between the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves includes:
[0023] Obtaining an average value of intermediate singular values of the same candidate position in the intermediate singular value matrix of several current sinusoidal periodic curves adjacent to the candidate current sinusoidal periodic curve; the candidate current sinusoidal periodic curve represents any current sinusoidal periodic curve; the candidate position represents any position in the intermediate singular value matrix;
[0024] The difference between the intermediate singular value of the candidate position of the candidate current sinusoidal periodic curve and the average value is taken as the difference between the corresponding intermediate singular value matrices of the candidate current sinusoidal periodic curve and several adjacent current sinusoidal periodic curves.
[0025] In combination with the first aspect above, in some possible implementations, the process of obtaining the target singular value matrix includes:
[0026] Acquire a first eigenvalue representing a fluctuation characteristic of an intermediate singular value matrix of the current sinusoidal periodic curve, and a second eigenvalue representing a difference between the intermediate singular value matrices corresponding to a plurality of adjacent current sinusoidal periodic curves;
[0027] fusing the first eigenvalue and the second eigenvalue to obtain an updated weight;
[0028] The intermediate singular value matrix is updated according to the update weights to obtain the target singular value matrix.
[0029] By combining the fluctuation characteristics of the intermediate singular value matrix of the current sinusoidal periodic curve and the difference between the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves, the intermediate singular value matrix is updated and corrected again, which can improve the accuracy of obtaining the target singular value matrix.
[0030] In combination with the first aspect above, in some possible implementations, denoising the output current data of the low-voltage circuit breaker circuit according to the target singular value matrix includes:
[0031] The target singular value matrix is replaced by the initial singular value matrix, and after restoration, the target local output current data is obtained.
[0032] In combination with the first aspect above, in some possible implementations, the low-voltage circuit breaker circuit fault detection method further includes:
[0033] Obtaining denoised target output current data according to the target local output current data of each time period;
[0034] The target output current data is compared with a preset current threshold, and if the preset current threshold is exceeded, an overload alarm signal is output.
[0035] In a second aspect, an embodiment of the present application provides a low-voltage circuit breaker circuit fault detection system, comprising a memory and a processor, wherein the memory is used to store executable program code, and the processor is used to call and run program code from the memory to implement the above-mentioned low-voltage circuit breaker circuit fault detection method.
[0036] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program code. When the computer program code runs on a computer, the computer executes the above-mentioned low-voltage circuit breaker circuit fault detection method. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] Figure 1 It is a flow chart of a method for detecting a fault in a low voltage circuit breaker circuit;
[0038] Figure 2 It is a schematic diagram of current sinusoidal curve fitting;
[0039] Figure 3 It is a schematic diagram of grid division;
[0040] Figure 4 It is a flowchart of the process of obtaining the intermediate singular value matrix;
[0041] Figure 5 It is a flowchart of the process of obtaining the target singular value matrix;
[0042] Figure 6 is a flow chart of other steps included in the low voltage circuit breaker circuit fault detection method;
[0043] Figure 7 It is a structural principle diagram of a low voltage circuit breaker circuit fault detection system;
[0044] Figure 8 It is a structural schematic diagram of a computer-readable storage medium. DETAILED DESCRIPTION
[0045] The technical solution in this application will be described clearly and in detail below in conjunction with the accompanying drawings.
[0046] This embodiment provides a low voltage circuit breaker circuit fault detection method, such as Figure 1 As shown, the following steps are included:
[0047] Step S1: obtaining output current data of a low voltage circuit breaker circuit and fitting it into a current sinusoidal curve, wherein the current sinusoidal curve includes current sinusoidal periodic curves of multiple time periods:
[0048] In the alternating current (AC) system, the direction of the current changes periodically, alternating between the positive and negative poles, which reduces the power loss of the AC when it is transmitted over long distances. The low-voltage circuit breaker is a controller and protection device in the AC system. In the low-voltage circuit breaker circuit, the voltage and current of the AC change in a sinusoidal waveform, and the output current also presents a sinusoidal waveform, that is, its amplitude and direction change in a sinusoidal manner over time.
[0049] In this embodiment, a current transformer is provided at the output end of the low-voltage circuit breaker circuit to detect the AC current output by the low-voltage circuit breaker circuit and obtain the output current data of the low-voltage circuit breaker circuit. The acquisition frequency is set according to actual needs, such as 1000 Hz. Therefore, the output current data of the low-voltage circuit breaker circuit includes the output current of the low-voltage circuit breaker circuit acquired according to the acquisition frequency.
[0050] Since the output current of the low-voltage circuit breaker circuit changes according to the sinusoidal variation law, the output current data of the low-voltage circuit breaker circuit is fitted into a current sinusoidal curve. It should be understood that the fitting algorithm used in the present embodiment when performing sinusoidal curve fitting is an existing sinusoidal curve fitting algorithm, for example, the output current data of the low-voltage circuit breaker circuit is input into MATLAB software, and the output current data of the low-voltage circuit breaker circuit is fitted using a sinusoidal function.
[0051] Since the sine curve has periodicity, that is, the current sine curve obtained by fitting is composed of curves of multiple time periods, and after the current sine curve is fitted, the boundary time point information of each time period can also be obtained, and the boundary time point information of each time period is the division time point of each time period. Figure 2 As shown, the current sinusoidal curve is placed in a two-dimensional coordinate system, the horizontal axis of the coordinate system is the time axis, and the vertical axis is the current value. As a specific example, the moment t0 in the horizontal axis is the time point of the current data collection start point, and the time points of each time period are t1, t2, t3 and t4. Therefore, the current sinusoidal curve is divided into time periods to obtain current sinusoidal periodic curves of multiple time periods, and each current sinusoidal periodic curve is a current sinusoidal curve within a sinusoidal period, that is, the current sinusoidal periodic curves are: the current sinusoidal curve between t0 and t1, the current sinusoidal curve between t1 and t2, the current sinusoidal curve between t2 and t3, and the current sinusoidal curve between t3 and t4.
[0052] Then, based on the division time points of each time period, the output current data of the low-voltage circuit breaker circuit is divided to obtain the local output current data of each time period. Therefore, the local output current data corresponds to the current sine cycle curve one by one. It should be understood that when performing curve fitting, not all the output current data of the low-voltage circuit breaker circuit are fitted into the curve, that is, not all the output current data of the low-voltage circuit breaker circuit are on the curve, and some of the output current data of the low-voltage circuit breaker circuit are outside the curve, such as above or below the curve, such as Figure 2 As shown, the data points above or below the current sinusoidal curve.
[0053] For the current data in the local output current data, the current data that is not on the current sinusoidal periodic curve is called discrete current data, that is, the current data that is not fitted into the current sinusoidal periodic curve is called discrete current data.
[0054] Step S2: According to the difference between the discrete current data outside the current sinusoidal periodic curve and the current sinusoidal periodic curve, the initial singular value matrix of the current data matrix after SVD decomposition is corrected to obtain an intermediate singular value matrix; the current data matrix is converted from the output current data of the low-voltage circuit breaker circuit in the time period corresponding to the current sinusoidal periodic curve:
[0055] In order to execute the subsequent SVD singular value decomposition algorithm, it is necessary to convert the local output current data of each time period into a current data matrix, and each local output current data of each time period corresponds to a current data matrix. Through matrix conversion, the local output current data of each time period can be converted into a current data matrix, so as to perform SVD singular value decomposition on the current data matrix.
[0056] In a specific embodiment, a specific process of obtaining the current data matrix is given as follows: the area of the local output current data in the two-dimensional coordinate system is grid-divided, such as Figure 3 As shown. The purpose of setting the grid is to include each current data of the local output current data into the corresponding grid. On the basis of meeting the above setting purpose, the grid size is set according to actual needs. As a specific implementation, the length of each grid on the horizontal axis can be set to a sampling period. For example, if the sampling period is 0.001s, the length of each grid on the horizontal axis is 0.001s; the length of each grid on the vertical axis can be determined according to the numerical value accuracy of the current data. For example, if the numerical value of the current data is taken to the unit digit, the length of each grid on the vertical axis can be 1A. If the numerical value of the current data is taken to 1 decimal place, the length of the grid on the vertical axis can be 0.1A. The data in each grid is used as each element of the current data matrix. The number of grids on the horizontal axis is set to N, and the number of grids on the vertical axis is set to M. Therefore, the current data matrix is an M×N data matrix.
[0057] Since the area of local output current data in the two-dimensional coordinate system is gridded, some areas have no current data and are blank areas, such as Figure 3As shown. The area with current data includes two situations, namely, the current data on the current sinusoidal curve and the current data that is not fitted into the current sinusoidal curve (i.e., discrete current data). Therefore, each current value of the area with current data is used as the data of the corresponding grid, and the value of the data of each grid in the blank area is assigned to a preset value. The preset value is set according to actual needs. In order to facilitate subsequent data processing, this embodiment determines the preset value to be 0. Therefore, the data of the grid with current data is set to the corresponding current data, and the value of the position without current data is set to 0, so as to achieve the assignment of all grids and obtain the current data matrix.
[0058] For ease of explanation, the candidate current sinusoidal periodic curve is assumed to represent any current sinusoidal periodic curve. In the following singular value correction and update process, the discrete current data all represent current data belonging to the same time period as the candidate current sinusoidal periodic curve.
[0059] The current data matrix is subjected to SVD singular value decomposition. The SVD singular value decomposition algorithm is an existing algorithm, and the implementation process of the algorithm itself is no longer described in detail. After the current data matrix is subjected to SVD singular value decomposition, three matrices will be obtained, namely: the transpose of the left singular vector matrix, the diagonal matrix, and the right singular vector matrix. Among them, the diagonal matrix is a singular value matrix, which is defined as the initial singular value matrix, and each singular value included is defined as the initial singular value. The diagonal matrix needs to be corrected and updated subsequently to achieve the update of the current data matrix, that is, the denoising of the current data.
[0060] After the current curve is fitted, for the candidate discrete current data outside the candidate current sinusoidal periodic curve, there will be a certain difference with the current sinusoidal periodic curve, and the candidate discrete current data is set to any discrete current data. The difference specifically refers to the difference between the candidate discrete current data and the fitting current data at the corresponding position in the current sinusoidal periodic curve. The corresponding position refers to the position on the current sinusoidal periodic curve with the same horizontal coordinate as the candidate discrete current data in the two-dimensional coordinate system. The reason for this difference is that after the curve is fitted, there will be a fitting current value corresponding to the candidate discrete current data on the current sinusoidal periodic curve, so the current difference between the candidate discrete current data and the corresponding fitting current value is obtained. The current difference is specifically the absolute value of the current difference between the candidate discrete current data and the corresponding fitting current value. The smaller the current difference between the candidate discrete current data and the corresponding fitting current value, the greater the possibility of belonging to normal current data, and vice versa.
[0061] The difference between the candidate discrete current data and the fitting current data at the corresponding position in the current sinusoidal periodic curve is used as the difference between the candidate discrete current data and the current sinusoidal periodic curve, so as to obtain the difference between each discrete current data and the current sinusoidal periodic curve. Then, according to the difference between each discrete current data and the current sinusoidal periodic curve, the initial singular value is corrected, and the corrected singular value is defined as the intermediate singular value. It should be understood that the singular values of the elements at different positions in the current data matrix after SVD decomposition are different, and the possibility that the current data closer to the current sinusoidal periodic curve belongs to noise is smaller, and the singular value should be larger, and the singular value farther from the current sinusoidal periodic curve should be smaller. Since the distance between the discrete current data and the current sinusoidal periodic curve is the difference between the discrete current data and the current sinusoidal periodic curve in the above, then, the greater the difference, the more it is necessary to adjust the singular value, so that the singular value farther from the current sinusoidal periodic curve should be smaller. Therefore, the correction coefficient is obtained according to the difference between each discrete current data and the current sinusoidal periodic curve, wherein the greater the difference between each discrete current data and the current sinusoidal periodic curve, the smaller the correction coefficient should be, and the two are in an inverse correlation. There are many ways to implement the inverse correlation, such as finding the inverse, or taking the difference as the power of a first preset value, the first preset value is a value greater than 0 and less than 1, or taking the opposite of the difference and then taking the opposite as the power of a second preset value, the second preset value is a value greater than 1, and so on. As a specific implementation, Figure 4 As shown, the correction process of the initial singular values is given as follows:
[0062] Step S201: Obtain the difference characteristic values between all discrete current data and the current sinusoidal periodic curve, the difference characteristic values are used to characterize the overall difference between all discrete current data and the current sinusoidal periodic curve, as a specific implementation, calculate the average value of the difference between all discrete current data and the current sinusoidal periodic curve, and use the average value as the overall difference. As another implementation, the sum of the differences between all discrete current data and the current sinusoidal periodic curve can also be used as the overall difference.
[0063] Step S202: Perform negative correlation normalization on the difference characteristic value to obtain a correction coefficient, that is, the difference characteristic value after negative correlation normalization is the correction coefficient. The negative correlation normalization is implemented as follows:
[0064]
[0065] Among them, y is the normalized value of negative correlation, x is the input value, and exp is the exponential function with the natural constant e as the base.
[0066] Through the above negative correlation normalization, the greater the difference, the smaller the weight.
[0067] Step S203: Using the correction coefficient as a weight, the initial singular value matrix is corrected to obtain an intermediate singular value matrix. Specifically, the initial singular value matrix is multiplied by the correction coefficient so that each initial singular value is multiplied by the correction coefficient. The resulting matrix is an intermediate singular value matrix, and each element on the diagonal of the intermediate singular value matrix is an intermediate singular value.
[0068] As a specific implementation, the calculation formula for obtaining the intermediate singular value matrix is given as follows:
[0069]
[0070] in, represents the initial singular value matrix, represents the initial singular value at the (1, 1)th position in the initial singular value matrix, represents the initial singular value at the (2, 2)th position in the initial singular value matrix; represents the intermediate singular value matrix, represents the intermediate singular value at the (1, 1)th position in the intermediate singular value matrix, represents the intermediate singular value at the (2, 2)th position in the intermediate singular value matrix; exp represents the exponential function with the natural constant e as the base; n represents the number of discrete current data, Represents the difference characteristic value between the i-th discrete current data and the current sinusoidal periodic curve.
[0071] Through the above process, the initial singular value matrix and the intermediate singular value matrix corresponding to the local output current data of each time period (ie, each current sinusoidal periodic curve) are obtained.
[0072] Step S3: Integrate the data fluctuation characteristics of the intermediate singular value matrix of the current sinusoidal periodic curve and the differences between the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves, update the intermediate singular value matrix, and obtain the target singular value matrix:
[0073] After step S2, the initial singular value matrix is corrected based on the acquired weights, but in the corrected intermediate singular value matrix, there are larger singular values that appear as normal data, and smaller singular values that appear as noise data, which cannot be uniformly updated and changed. Therefore, it is also necessary to update and correct the intermediate singular values again according to the data fluctuation characteristics of the intermediate singular value matrix of the current sinusoidal periodic curve. Moreover, there is also a certain correlation between the current sinusoidal periodic curve and other adjacent current sinusoidal periodic curves. In order to ensure the accuracy of the update of the intermediate singular value, it is also necessary to combine the intermediate singular value matrix of the current sinusoidal periodic curve with the difference between the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves to update the intermediate singular value matrix. Therefore, it is necessary to fuse the data fluctuation characteristics of the intermediate singular value matrix of the current sinusoidal periodic curve and the difference between the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves, update the intermediate singular value matrix, and obtain the target singular value matrix.
[0074] Among them, the data fluctuation characteristics of the intermediate singular value matrix of the current sinusoidal periodic curve are used to characterize the data fluctuation of the intermediate singular values in the intermediate singular value matrix of the current sinusoidal periodic curve. It should be understood that the relevant parameters used to characterize the data fluctuation characteristics can be used for calculation, such as variance. The data fluctuation characteristics of the intermediate singular value matrix can reflect the abnormality of the intermediate singular value. The more abnormal, the more it needs to correct and update the intermediate singular value, that is, the larger the correction amplitude should be.
[0075] The difference between the intermediate singular value matrix of the current sinusoidal periodic curve and the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves is used to characterize the difference between the intermediate singular value matrix of the current sinusoidal periodic curve and the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves. The greater the difference, the lower the similarity. Because the current data is a sinusoidal periodic change, there should be a certain similarity between the intermediate singular value matrix of the current sinusoidal periodic curve and the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves. The more similar, the more normal the current data corresponding to the corresponding intermediate singular value is, and the less the intermediate singular value needs to be corrected and updated, that is, the smaller the correction amplitude should be; the less similar, the more abnormal the current data corresponding to the corresponding intermediate singular value is, and the more the intermediate singular value needs to be corrected and updated, that is, the larger the correction amplitude should be.
[0076] As a specific implementation, the adjacent plurality of current sinusoidal periodic curves of the candidate current sinusoidal periodic curve are a set number of current sinusoidal periodic curves adjacent to the candidate current sinusoidal periodic curve on the left and right, that is, a set number of current sinusoidal periodic curves closest to the candidate current sinusoidal periodic curve. The set number is set according to actual needs, such as 10 adjacent current sinusoidal periodic curves on the left and right. Left and right adjacent means left adjacent and / or right adjacent, that is, if there are at least five current sinusoidal periodic curves on the left and right sides of the candidate current sinusoidal periodic curve, then the 5 left adjacent and 5 right adjacent current sinusoidal periodic curves can be selected respectively as the 10 left and right adjacent current sinusoidal periodic curves. If there are less than five current sinusoidal periodic curves on the left or right side of the candidate current sinusoidal periodic curve, it is necessary to select more from the other side to make up 10 current sinusoidal periodic curves. For example: if there is only one current sinusoidal periodic curve adjacent to the left of the candidate current sinusoidal periodic curve, it is necessary to select 9 from the right adjacent to make up 10 current sinusoidal periodic curves. Or if the candidate current sinusoidal periodic curve is the starting current sinusoidal periodic curve and there is no current sinusoidal periodic curve adjacent to the left, then 10 current sinusoidal periodic curves adjacent to the right are required as the 10 adjacent current sinusoidal periodic curves.
[0077] As a specific implementation, the process of obtaining the difference between the intermediate singular value matrix of the candidate current sinusoidal periodic curve and the corresponding intermediate singular value matrix of several adjacent current sinusoidal periodic curves includes: setting the candidate position to represent any position in the intermediate singular value matrix, since each current sinusoidal periodic curve corresponds to an intermediate singular value matrix, therefore, obtaining the intermediate singular value of the same candidate position in the intermediate singular value matrix of several current sinusoidal periodic curves adjacent to the candidate current sinusoidal periodic curve, and calculating the average value of the intermediate singular value of the same candidate position. The average value can reflect the overall level of the intermediate singular value of the candidate position corresponding to the several adjacent current sinusoidal periodic curves, then, the difference between the intermediate singular value of the candidate position of the candidate current sinusoidal periodic curve and the average value is used as the difference between the candidate current sinusoidal periodic curve and the corresponding intermediate singular value of the several adjacent current sinusoidal periodic curves, thereby obtaining the difference corresponding to the intermediate singular value of each position in the intermediate singular value matrix of the candidate current sinusoidal periodic curve. In this embodiment, the difference is specifically the absolute value of the difference.
[0078] In one embodiment, Figure 5 As shown in FIG. 1 , the process of obtaining the target singular value matrix includes:
[0079] Step S301: Obtain a first eigenvalue representing the fluctuation characteristics of the intermediate singular value matrix of the candidate current sinusoidal periodic curve, and a second eigenvalue representing the difference between the candidate current sinusoidal periodic curve and the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves. The first eigenvalue may be specifically a variance, and the second eigenvalue may be a difference corresponding to the intermediate singular value of each position in the intermediate singular value matrix of the candidate current sinusoidal periodic curve.
[0080] Step S302: The first eigenvalue and the second eigenvalue are integrated to obtain an update weight corresponding to each intermediate singular value in the intermediate singular value matrix of the candidate current sinusoidal periodic curve.
[0081] Step S303: updating each intermediate singular value in the intermediate singular value matrix according to the update weight to obtain the corresponding target singular value, and all the target singular values constitute the target singular value matrix.
[0082] Among them, since the first eigenvalue represents the fluctuation of the intermediate singular value matrix of the candidate current sinusoidal periodic curve, the greater the fluctuation, the more it is necessary to correct and update the intermediate singular value, and the second eigenvalue represents the difference between the corresponding intermediate singular value matrices of the candidate current sinusoidal periodic curve and several adjacent current sinusoidal periodic curves. The greater the difference, the more it is necessary to correct and update the intermediate singular value. Therefore, the correlation between the first eigenvalue and the second eigenvalue and the update weight is the same, then, when the first eigenvalue and the second eigenvalue are fused, the first eigenvalue and the second eigenvalue can be added or multiplied. In a specific embodiment, the first eigenvalue and the second eigenvalue are multiplied, and the obtained product is normalized to the maximum and minimum values to obtain the initial update weight. In addition, in order to further ensure that the larger intermediate singular value is larger after the update, the above initial update weight can be added to a hyperparameter, and the specific value of the hyperparameter is set according to actual needs, such as 0.2, which is used to make the initial update weight with a larger value plus the hyperparameter greater than 1, so as to ensure that the larger intermediate singular value is larger after the update.
[0083] According to the update weight, each intermediate singular value in the intermediate singular value matrix is updated to obtain the corresponding target singular value, specifically: each intermediate singular value in the intermediate singular value matrix is multiplied by the corresponding update weight, and the result obtained is the target singular value.
[0084] As a specific implementation, the calculation formula for obtaining the target singular value is given as follows:
[0085]
[0086] in, represents the target singular value at the (j, j)th position on the target singular value matrix, f[] represents the maximum and minimum value normalization function, and m represents the number of intermediate singular values in the intermediate singular value matrix, that is, (the number of intermediate singular values on the diagonal matrix corresponding to the intermediate singular value matrix); represents the intermediate singular value at the (l, l)th position in the intermediate singular value matrix, represents the average of the intermediate singular values of the intermediate singular value matrix, represents the intermediate singular value at the (j, j)th position on the intermediate singular value matrix, It represents the mean value of the intermediate singular value at the (j, j)th position in the intermediate singular value matrix corresponding to 10 adjacent current sinusoidal periodic curves; represents a hyperparameter.
[0087] Through the above process, the target singular value matrix corresponding to each current sinusoidal periodic curve can be obtained.
[0088] Step S4: De-noising the output current data of the low voltage circuit breaker circuit according to the target singular value matrix.
[0089] After obtaining the target singular value matrix corresponding to each current sinusoidal periodic curve, the output current data of the low-voltage circuit breaker circuit can be denoised according to the target singular values of the target singular value matrix corresponding to each current sinusoidal periodic curve. In this embodiment, for the candidate current sinusoidal periodic curve, the target singular value matrix of the candidate current sinusoidal periodic curve replaces the initial singular value matrix of the candidate current sinusoidal periodic curve, and after restoration, the target local output current data is obtained. Among them, as described above, after the SVD singular value decomposition of the current data matrix, three matrices will be obtained, namely: the transposition of the left singular vector matrix, the diagonal matrix and the right singular vector matrix. Therefore, the target singular value matrix replaces the initial singular value matrix, and combined with the transposition of the left singular vector matrix and the right singular vector matrix, the reverse data processing of the SVD singular value decomposition is performed, and the current data matrix corresponding to the candidate current sinusoidal periodic curve can be obtained, thereby obtaining the local output current data, which is defined as the target local output current data, and the data denoising process is realized.
[0090] In this embodiment, after data denoising, Figure 6 As shown, a low voltage circuit breaker circuit fault detection method also includes the following process:
[0091] Step S5: Obtain denoised target output current data according to the target local output current data of each time period. Specifically, the target local output current data of each time period are combined according to the time sequence to obtain denoised target output current data.
[0092] Step S6: Compare the target output current data with the preset current threshold. If it exceeds the preset current threshold, an overload alarm signal is output. The preset current threshold is set according to actual needs. Each current data in the target output current data is compared with the preset current threshold. If current data exceeding the current threshold appears, it indicates that the current is overloaded, and an overload alarm signal is output. As a specific implementation, the overload alarm signal can be output to an alarm. As other implementations, two current thresholds can be set, namely an upper current threshold and a lower current threshold. If the current data in the target output current data is greater than the upper current threshold, it is determined to be an overload alarm signal; if the current data in the target output current data is less than the lower current threshold, there may be a situation where the power supply voltage is insufficient or the power line contact is poor, and the output current is too low. Alarm signal.
[0093] This embodiment also provides a low voltage circuit breaker circuit fault detection system 700, such as Figure 7 As shown, it includes a memory 701 and a processor 702, the memory 701 is used to store an executable program code 703, and the processor 702 is used to call and run the program code 703 from the memory 701 to implement Figure 1 A method for detecting circuit faults in a low voltage circuit breaker is shown.
[0094] This embodiment also provides a computer-readable storage medium 800, such as Figure 8 As shown, the computer readable storage medium 800 stores a computer program code 801. When the computer program code 801 is executed on a computer, the computer executes Figure 1 A method for detecting circuit faults in a low voltage circuit breaker is shown.
[0095] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.
Claims
1. A method for detecting a circuit fault of a low voltage circuit breaker, characterized in that: include: Acquire output current data of a low voltage circuit breaker circuit, and fit it into a current sinusoidal curve, wherein the current sinusoidal curve includes current sinusoidal periodic curves of multiple time periods; According to the difference between the discrete current data outside the current sinusoidal periodic curve and the current sinusoidal periodic curve, the initial singular value matrix of the current data matrix after SVD decomposition is corrected to obtain an intermediate singular value matrix; the current data matrix is converted from the output current data of the low-voltage circuit breaker circuit in the time period corresponding to the current sinusoidal periodic curve; The data fluctuation characteristics of the intermediate singular value matrix of the current sinusoidal periodic curve are integrated, and the differences between the intermediate singular value matrices corresponding to several adjacent current sinusoidal periodic curves are updated to obtain the target singular value matrix; De-noising the output current data of the low voltage circuit breaker circuit according to the target singular value matrix; The process of obtaining the intermediate singular value matrix includes: Obtaining difference characteristic values between all discrete current data corresponding to the current sinusoidal periodic curve and the current sinusoidal periodic curve, wherein the difference characteristic values are used to characterize the overall difference between all discrete current data and the current sinusoidal periodic curve; Performing negative correlation normalization on the difference characteristic value to obtain a correction coefficient; Using the correction coefficients as weights, the initial singular value matrix is corrected to obtain an intermediate singular value matrix; The process of acquiring the current data matrix includes: Divide the area of the local output current data in the two-dimensional coordinate system into grids, and use the data in each grid as the elements of the current data matrix, wherein the local output current data is used as the data in the corresponding grid, and the data in other grids are assigned to preset values; the local output current data is the output current data of the low-voltage circuit breaker circuit in the time period corresponding to the current sine period curve; The process of obtaining the target singular value matrix includes: Acquire a first eigenvalue representing a fluctuation characteristic of an intermediate singular value matrix of the current sinusoidal periodic curve, and a second eigenvalue representing a difference between the intermediate singular value matrices corresponding to a plurality of adjacent current sinusoidal periodic curves; fusing the first eigenvalue and the second eigenvalue to obtain an updated weight; The intermediate singular value matrix is updated according to the update weights to obtain the target singular value matrix.
2. A low voltage circuit breaker circuit fault detection method according to claim 1, characterized in that: The process of obtaining the difference between discrete current data outside the current sinusoidal periodic curve and the current sinusoidal periodic curve includes: The difference between the candidate discrete current data and the fitting current data at the corresponding position in the current sinusoidal periodic curve is taken as the difference between the candidate discrete current data and the current sinusoidal periodic curve, wherein the candidate discrete current data is any discrete current data.
3. A low voltage circuit breaker circuit fault detection method according to claim 1, characterized in that: The process of obtaining the difference between the corresponding intermediate singular value matrices of several adjacent current sinusoidal periodic curves includes: Obtaining an average value of intermediate singular values of the same candidate position in the intermediate singular value matrix of several current sinusoidal periodic curves adjacent to the candidate current sinusoidal periodic curve; the candidate current sinusoidal periodic curve represents any current sinusoidal periodic curve; the candidate position represents any position in the intermediate singular value matrix; The difference between the intermediate singular value of the candidate position of the candidate current sinusoidal periodic curve and the average value is taken as the difference between the corresponding intermediate singular value matrices of the candidate current sinusoidal periodic curve and several adjacent current sinusoidal periodic curves.
4. A low voltage circuit breaker circuit fault detection method according to claim 1, characterized in that: According to the target singular value matrix, the output current data of the low voltage circuit breaker circuit is denoised, including: The target singular value matrix is replaced by the initial singular value matrix, and after restoration, the target local output current data is obtained.
5. A low voltage circuit breaker circuit fault detection method according to claim 4, characterized in that: The low voltage circuit breaker circuit fault detection method further comprises: Obtaining denoised target output current data according to the target local output current data of each time period; The target output current data is compared with a preset current threshold, and if the preset current threshold is exceeded, an overload alarm signal is output.
6. A low voltage circuit breaker circuit fault detection system, characterized in that: It comprises a memory and a processor, wherein the memory is used to store executable program code, and the processor is used to call and run program code from the memory to implement a low-voltage circuit breaker circuit fault detection method as described in any one of claims 1-5.
7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program codes, and when the computer program codes are executed on a computer, the computer is enabled to execute a low-voltage circuit breaker circuit fault detection method as claimed in any one of claims 1 to 5.
