A method for synchronizing fault signals in a power system

Abstract

This invention discloses a method for synchronizing power system fault signals, comprising the following steps: (1) acquiring three-phase current data collected by each waveform recorder during a power system fault, and extracting wavelet detail coefficients of each phase current data using wavelet transform; (2) calculating the minimum intersection of the fault waveform recording data of each waveform recorder before and after the signal abrupt change; (3) sequentially setting the time abscissa from 0 according to the corresponding sampling interval within the minimum intersection time of each phase, acquiring the sampling data of each waveform recorder within the common minimum intersection time, and completing the synchronization of the fault signals of each waveform recorder. Compared with the current use of clock synchronization or manual identification and analysis, this invention utilizes wavelet transform to more accurately detect the abrupt change point of the fault, reducing labor costs and improving efficiency and accuracy.

Description

Technical Field

[0001] This invention belongs to the field of power system fault technology, and specifically relates to a method for synchronizing power system fault signals. Background Technology

[0002] To meet the needs of the national economy, the scale of the power grid has been increasing year by year, leading to gradually higher voltage levels and increasingly complex transmission lines. Therefore, power quality has become a crucial issue affecting national production and daily life. Although many international standards have been established for solving power problems, the detection of power system faults remains increasingly difficult due to their complexity. Currently, when a power grid fault occurs, fault recording devices are typically used to record transient data. However, due to the inconsistent standards of current fault recording devices, even different models from the same manufacturer can vary in accuracy. Furthermore, different recording devices use different standards for recording the same fault, leading to significant errors in data recording and subsequent fault diagnosis. Therefore, studying the information recorded by different recorders and then filtering out identical fault data transmitted from different recorders for subsequent fault data synchronization has significant theoretical and practical value.

[0003] When a power system fault occurs, a large amount of data is recorded by fault recorders, providing decision support for grid dispatchers to analyze and handle the fault and restore normal system operation. Currently, fault information analysis and processing often requires fault waveform data from both sides of the line, and the sampling data needs to be taken at the same time. However, due to time differences or asynchronous start-up times of the devices on both sides of the line, the obtained waveform data is not synchronized, which hinders subsequent fault analysis. Therefore, when synchronizing waveform files acquired by different recorders, clock synchronization is often used to achieve waveform data alignment. This requires the recorder to obtain the time deviation relative to the master station and correct the time deviation of the recorder. However, the disadvantages of achieving waveform data alignment through time synchronization are low reliability and susceptibility to external factors, such as the large electromagnetic interference process that often accompanies the initial stage of a fault and the discrete deviation of the recorder sampling, which affect the timely analysis and processing of fault information.

[0004] Chinese Patent Publication No. CN 110535551 A discloses a method and system for synchronizing fault waveform sampling data in a power system. The method includes: analyzing and storing fault waveform sampling SV message data with hardware absolute time stamps from each merging unit of a substation; selecting a reference merging unit and using its absolute time stamp as the absolute time stamp for synchronization; finding the interpolation position based on the absolute time stamp of the synchronization time, calculating the interpolated data for the synchronization time, and completing the synchronization interpolation of the waveform sampling data. This patent does not rely on a station-wide synchronization signal, solving the problem of limited FPGA synchronization data computation resources when numerous SVs are connected. It achieves synchronization of power system fault waveform sampling data in non-real-time systems, improving the accuracy and amplitude consistency of the sampling data, and has good application prospects. However, the synchronization method of this patent cannot eliminate the discrete deviation of the waveform recorder sampling at the beginning of a power fault and electromagnetic interference, which can easily cause errors in the synchronization data. Chinese Patent Publication No. CN1312472A discloses a method and device for synchronously recording transient changes in a power system. A high-precision crystal oscillator drives AD sampling to cyclically acquire measurement signals. When transient processes such as faults or switching operations occur, the precise time of the start signal is recorded, and the acquisition results before and after the start are saved. The clock is implemented using a GPS-synchronized crystal oscillator, eliminating deviations caused by SA interference and satellite loss of lock, with a self-generated second pulse deviation of less than 0.1 μS. Under the synchronization of the clock, multiple recording devices synchronously measure signals from remote measurement points, connecting to form a network to achieve distributed high-speed recording and measurement of transient changes such as power system operation, faults, and disturbances. However, the synchronous data of this patent has low reliability and is easily affected by external factors, which can easily lead to significant errors in the subsequent judgment of power system faults. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a method for synchronizing power system fault signals. This synchronization method solves the problems of low reliability and accuracy when using clock synchronization methods to synchronize data in existing technologies.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for synchronizing fault signals in a power system includes the following steps:

[0008] (1) Obtain the three-phase current data collected by each waveform recorder when the power system is faulty, and use wavelet transform to extract the wavelet detail coefficients of each phase current data of each waveform recorder.

[0009] (2) Calculate the minimum intersection of the time windows of the fault recording data of each recorder before and after the moment of signal change;

[0010] (3) Set the time axis sequentially from 0 according to the corresponding sampling interval during the minimum intersection time of each phase. The point value of each time axis corresponding to each waveform recorder is the sampling value of the corresponding number of sampling points before and after the signal change time of the waveform recorder. Obtain the sampling data of each waveform recorder during the common minimum intersection time and complete the synchronization of the fault signals of each waveform recorder.

[0011] Furthermore, the specific process of extracting wavelet detail coefficients of each phase current data in step (1) is as follows: First, the three-phase current sampling data are divided into different column matrices by phase; second, the center point of the filter array is respectively assigned to each point of the matrix, and the data in front of the center point of the wavelet function array is padded with zeros; finally, the corresponding sampled value array and the filter array are inner products to obtain the detail coefficients corresponding to each point of the matrix, thereby obtaining the first layer wavelet detail coefficients of each phase current data.

[0012] Furthermore, the wavelet transform in step (1) uses the db3 wavelet function.

[0013] Furthermore, in step (2), the signal abrupt change time is the time corresponding to when the wavelet detail coefficients are greater than the threshold value.

[0014] Furthermore, the specific process of calculating the minimum intersection in step (2) is as follows: First, multiply the ordinal number of the sampling point corresponding to the sudden change time by the sampling interval to obtain the data time window of each recorder before and after the sudden change time; then find the intersection of the data time windows of each recorder to obtain the minimum intersection.

[0015] Furthermore, the specific process of synchronizing the fault signals of each waveform recorder in step (3) is as follows: First, divide the length of the minimum intersection time by the sampling interval to obtain the number of time abscissa points. Starting from the origin of the abscissa, each additional sampling interval yields a time abscissa point. Second, assign the sampling values ​​of the same number of sampling points before and after the signal change moment of each waveform recorder fault waveform data to each time abscissa point, thus obtaining the waveform within the time window in which each waveform recorder completes synchronization.

[0016] Furthermore, the minimum intersection time is the time window during which each waveform recorder needs to be synchronized.

[0017] Compared with the prior art, the positive and beneficial effects of this invention are as follows:

[0018] Compared to current methods that rely on clock synchronization or manual analysis, this invention acquires three-phase current data collected by various recorders during power system faults. It then uses wavelet transform to extract wavelet detail coefficients from the current data of each phase, calculates the minimum intersection of the recorder's fault recording data within the time windows before and after the signal abrupt change, and sequentially sets time axes starting from 0 according to the corresponding sampling intervals within each phase's minimum intersection time. The point value corresponding to each time axis for each recorder represents the sampled value of the corresponding number of sampling points before and after the signal abrupt change for that recorder. This yields the sampled data from all recorders within the shared minimum intersection time, thus synchronizing the fault signals of each recorder. This invention, utilizing wavelet transform, can more accurately detect fault abrupt changes, reduces labor costs, improves efficiency, and consequently enhances the accuracy of power system fault detection. Attached Figure Description

[0019] Figure 1 This is a diagram of the Mallat wavelet computation decomposition process in this invention;

[0020] Figure 2 This is a diagram showing the total amount of addition and multiplication calculations required for calculating the wavelet detail coefficients at each layer in this invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0022] Example

[0023] A method for synchronizing fault signals in a power system is disclosed. This method utilizes different fault recording devices to record electrical signals at the time of a fault, storing these signals as a common electrical signal in the recording files of their respective devices. By reading the recording files from different fault recording devices at the same fault occurrence, the waveforms of the three-phase currents (phase A, phase B, and phase C) are extracted from the recording files. The extracted three-phase current values ​​are stored as a single matrix (N x 1). Wavelet transform is then performed on this matrix to obtain wavelet detail coefficients. The maximum absolute value of the wavelet detail coefficients is compared with a threshold value set for abrupt change detection to detect abrupt changes in the current of each phase. Based on this, the minimum intersection of the time windows of the recorded data before and after the abrupt change point is obtained to achieve signal synchronization between the fault recording devices in the power plant. The specific process of this method is as follows:

[0024] (1) Write a program to read the waveform recording files of each fault recorder when the same fault occurs, and obtain all the signal quantities collected by each fault recorder in the plant. The waveform recording files contain switch action information, trip information, and reclosing information.

[0025] (2) Select the three-phase current waveforms of phase A, phase B and phase C collected by each fault recorder in the plant, obtain the sampling data of each phase of the three-phase current, and form different matrices for each data. Therefore, the three-phase current contained in each fault recorder will form 3 N rows and 1 columns matrices.

[0026] (3) The Mallat wavelet algorithm is applied to the three matrices formed by the three-phase currents acquired by the fault recorder. This wavelet transform uses decimation wavelet calculation, which is essentially the process of the inner product of the sampling points and the filter array. It is the result of a series of addition and multiplication operations, that is, extracting the high-frequency components of the fault. Taking the db3 wavelet as an example, the process of calculating the wavelet coefficients of the first layer of the wavelet algorithm is as follows: Figure 1 As shown.

[0027] Based on this, and combining the Mallat algorithm's computational process, the total number of additions and multiplications required for calculating wavelet coefficients at each level can be obtained, such as... Figure 2 As shown, n is the number of sampling points in the data window selected by the protection algorithm, and the calculation result of the detail coefficients depends on the parity of n. Taking an odd number of n as an example, the number of first-layer smoothing approximation coefficients is (n+1) / 2, so it can be seen that obtaining the first-layer smoothing approximation coefficients requires 3(n+1) multiplications and 2.5(n+1) additions. For the calculation of the second-layer wavelet coefficients, it is necessary to determine the parity of (n+1) / 2. When (n+1) / 2 is even, the number of second-layer detail coefficients is (n+1) / 4, so it can be seen that obtaining the second-layer detail coefficients requires 1.5(n+1) multiplications and 1.25(n+1) additions. In summary, obtaining the second-layer wavelet detail coefficients requires a total of 4.5(n+1) multiplications and 3.75(n+1) additions.

[0028] Using the db3 wavelet function as an example, taking phase A current as an example, wavelet transform is used to extract the first-level detail coefficients and smooth approximation coefficients of phase A current. The specific process is as follows: First, the center point of the filter array is mapped to the first point of the matrix composed of phase A currents, and zeros are padded to the data before the center point of the db3 wavelet function array. Second, the inner product of the constructed sampled value array and the filter array is performed to obtain the smooth approximation coefficients or detail coefficients corresponding to the first point of the matrix composed of phase A currents. Then, the center point of the filter array is mapped to the second point of the matrix composed of phase A currents, and the inner product of the sampled signal and the filter array is performed to obtain the smooth approximation coefficients or detail coefficients corresponding to the second point of the matrix composed of phase A currents. Finally, this process is repeated until the smooth approximation coefficients or detail coefficients corresponding to all points of the matrix composed of phase A currents are obtained, which is the result of the first-level wavelet transform decomposition of the matrix composed of phase A currents.

[0029] It should be emphasized that when performing a wavelet transform on the three-phase current to obtain the corresponding detail coefficients, the maximum value of the absolute value of the wavelet detail coefficients when the system is running normally will be found by setting the threshold value of the abrupt change detection criterion.

[0030] (4) If the value of the wavelet detail coefficient is greater than the threshold value, then the time corresponding to the value is considered to be the signal change time, and recorded as follows: For example, the change point time of a certain element of the matrix composed of phase A current is t1, and with this time as the reference value, the minimum intersection of the time windows of each recorded wave data before the change time and the minimum intersection of the time windows of each recorded wave data after the change time are obtained respectively; The specific calculation method is as follows: The time corresponding to the fault sampling point (i.e. the signal change time point): Let index_X be the corresponding sampling point (the time corresponding to the fault sampling point), and the first layer detail coefficient is represented in this way:

[0031] —1—2—3—4—5—6—7;

[0032] Assuming the 5th sampling point is index_X, then according to the above representation, there are 5 sampling intervals "-" before it. The sampling interval is the reciprocal of the sampling frequency, and therefore the corresponding time is:

[0033] t=index_X×Ts=index_X÷fs;

[0034] Where Ts is the corresponding sampling interval and fs is the corresponding sampling frequency.

[0035] The minimum intersection tI min of the time windows of each waveform data before the abrupt change:

[0036] First, use the formula t = index_X × Ts = index_X ÷ fs to obtain the data window of each fault recorder before the abrupt change:

[0037] Fault recorder ①: (index_X-1) / fs1;

[0038] Fault recorder ②: (index_Y-1) / fs2;

[0039] Fault recorder ③: (index_Z-1) / fs3;

[0040] index_X, index_Y, and index_Z are the ordinal numbers of the fault sampling points of the corresponding waveform recorders in the data;

[0041] Then find the intersection, that is, take the minimum value t2 among the three:

[0042] t2=tII_min=min(min((size(x,1)-index_X) / fs1,(size(y,1)-index_Y) / fs2),(size(z,1)-index_Z) / fs3);

[0043] (5) For each fault recorder, extract the sampling data of the time window before and after the sudden change moment; in step (4), the total length of the common time window of the three signals that need to be synchronized is t1+t2. Just extract the sampling data of the common time window from the total time window of each of the three fault recorders. Take the A phase current as an example.

[0044] First, determine the horizontal coordinate corresponding to the segment being intercepted. The length of the time window is t1+t2. Then, within this time period, several points can be taken at the sampling frequency of the A-phase current. The calculation formula is floor((t1+t2)÷Ts1), where floor represents rounding down. The total time corresponding to these points is floor((t1+t2)÷Ts1) multiplied by a sampling interval Ts1, which is the horizontal coordinate range of the intercepted segment. When generating the waveform, we can start from the origin 0 of the horizontal coordinate and add a sampling interval Ts1 in each loop as the horizontal coordinate of the next point to be assigned (corresponding to the time horizontal coordinate in the invention content), and so on.

[0045] Then, determine the vertical coordinate corresponding to the extracted portion, which corresponds to the sampled value of each recorder at the corresponding time. The minimum intersection of the time windows of each recorder data before the sudden change is t1. During this time, the A-phase current is sampled at intervals of Ts1. A total of t1÷Ts1 points can be sampled within this interval. The first point is assigned a value using a for loop. The first point is the sampled value of the fault point with the ordinal number index_X-(t1÷Ts1). That is, for the number of points that need to be assigned within time t1, count several sampling points backward from the fault point index_X and assign the sampled values ​​of these sampling points to the time axis points that need to be assigned in chronological order. The same method is used for time t2. That is, calculate the number of points that need to be assigned according to the sampling frequency within time t2, count several sampling points backward from the fault point index_X, and assign the sampled values ​​of these sampling points to the time axis points that need to be assigned in chronological order. The time axis corresponding to the fault point is assigned by the fault sampling point.

[0046] (6) Extract and plot the sampling data of the time abscissa points that need to be assigned values ​​within the time window t1+t2 before and after all the sudden change times obtained by each waveform recorder.

[0047] (7) After the sudden change waveform is drawn, the fault recording data of each waveform recorder is synchronized and then it can be used for subsequent fault transient analysis.

[0048] The specific embodiments of the present invention have been given above, but the present invention is not limited to the described embodiments. Under the concept given by the present invention, the technical means in the above embodiments can be changed, replaced, or modified in a way that is easy for those skilled in the art to conceive of, and the effect is basically the same as the corresponding technical means in the present invention, and the purpose of the invention is also basically the same. The technical solution formed in this way is a fine-tuning of the above embodiments, and such technical solution still falls within the protection scope of the present invention.

Claims

1. A method for synchronizing fault signals in a power system, characterized in that, Includes the following steps: (1) Obtain the three-phase current data collected by each waveform recorder when the power system is faulty, and use wavelet transform to extract the wavelet detail coefficients of each phase current data of each waveform recorder. (2) Calculate the minimum intersection of the time windows of the fault recording data of each recorder before and after the moment of signal change; (3) Set the time axis sequentially from 0 according to the corresponding sampling interval during the minimum intersection time of each phase. The point value of each time axis corresponding to each waveform recorder is the sampling value of the corresponding number of sampling points before and after the signal change time of the waveform recorder. Obtain the sampling data of each waveform recorder during the common minimum intersection time and complete the synchronization of the fault signals of each waveform recorder. The specific process of synchronizing the fault signals of each waveform recorder in step (3) is as follows: First, divide the length of the minimum intersection time by the sampling interval to obtain the number of time abscissa points. Starting from the origin of the abscissa, each additional sampling interval will result in a time abscissa point. Second, assign the sampling values ​​of the same number of sampling points before and after the signal change moment of each waveform recorder fault waveform data to each time abscissa point, so as to obtain the waveform within the time window in which each waveform recorder completes synchronization.

2. The method for synchronizing power system fault signals according to claim 1, characterized in that, The specific process for extracting wavelet detail coefficients of each phase current data in step (1) is as follows: First, the three-phase current sampling data are divided into different column matrices by phase; second, the center point of the filter array is respectively assigned to each point of the matrix, and the data in front of the center point of the wavelet function array is padded with zeros; finally, the corresponding sampled value array and the filter array are inner products to obtain the detail coefficients corresponding to each point of the matrix, thereby obtaining the first layer wavelet detail coefficients of each phase current data.

3. The method for synchronizing power system fault signals according to claim 1, characterized in that, The wavelet transform in step (1) uses the db3 wavelet function.

4. The method for synchronizing power system fault signals according to claim 1, characterized in that, In step (2), the signal abrupt change time is the time when the wavelet detail coefficient is greater than the threshold value.

5. The method for synchronizing power system fault signals according to claim 1, characterized in that, The specific process of calculating the minimum intersection in step (2) is as follows: First, multiply the ordinal number of the sampling point corresponding to the sudden change time by the sampling interval to obtain the data time window of each recorder before and after the sudden change time; then find the intersection of the data time windows of each recorder to obtain the minimum intersection.

6. The method for synchronizing power system fault signals according to claim 1, characterized in that, The minimum intersection time is the time window during which each waveform recorder needs to be synchronized.

Images