A TEV pulse sampling and counting method

Abstract

The present invention discloses a TEV pulse sampling and counting method, which relates to the technical field of pulse sampling and counting methods, and includes the following steps: collecting TEV signals under undersampling conditions, with the sampling rate set to 250K; downsampling the collected TEV data to 5000 data points; based on a PRPD diagram drawn by TEV data, clustering the data points in the PRPD diagram drawn by TEV data by a DBSCAN algorithm, one category is background noise, and the other category is potential partial discharge points, eliminating the data points of background noise and retaining the potential partial discharge points. The present invention solves the problem of how to realize the statistics of the number of TEV pulses under undersampling conditions, which is of great significance to the miniaturization and practical application of equipment based on the TEV detection method, and significantly improves the accuracy and comprehensiveness of interference point removal, so that the number of TEV pulses can be better counted.

Description

Technical Field

[0001] The invention relates to the technical field of pulse sampling and counting methods, and in particular to a TEV pulse sampling and counting method. Background Art

[0002] DBSCAN is a density-based clustering algorithm. Compared with the traditional K-Means algorithm, DBSCAN can find clusters of any shape, and can also find outliers while clustering. It has a better clustering effect for dense data sets. In the TEV-based partial discharge detection method, TEV counting is an important reference, and the severity of partial discharge can be intuitively judged according to the number of pulses. However, since the frequency band of TEV signals is distributed within 3-100MHZ, it is difficult to take into account both distortion-free sampling and smaller sampling equipment. Therefore, the current mainstream approach is to collect TEV signals by undersampling, but this will cause various mature signal analysis methods to fail and be difficult to use. How to accurately count the number of TEV pulses under undersampling has become a major problem.

[0003] In the sinusoidal voltage waveform, 90 degrees and 270 degrees represent the instants when the voltage waveform changes from a negative peak to a positive peak (or vice versa). Near these two phase points, the instantaneous value of the voltage changes rapidly, which easily leads to the concentration and enhancement of the local electric field in the insulation system, thereby inducing partial discharge, that is, partial discharge usually occurs near 90 degrees or 270 degrees. Therefore, when partial discharge exists, it can be clearly found from the PRPD graph that data points are clustered near a certain phase. Therefore, the density-based DBSCAN algorithm can better distinguish between partial discharge points and background noise data points. Further analysis and processing on this basis can better count the number of TEV pulses. Summary of the invention

[0004] The object of the present invention is to provide a TEV pulse sampling and counting method to solve the above-mentioned deficiencies in the prior art.

[0005] In order to achieve the above object, the present invention provides the following technical solution: a TEV pulse sampling and counting method, comprising the following steps:

[0006] Step 1: Collect TEV signal under undersampling condition, and set the sampling rate to 250K;

[0007] Step 2: Downsample the collected TEV data to 5000 data points;

[0008] Step 3: Based on the PRPD diagram drawn by TEV data, the data points in the PRPD diagram drawn by TEV data are clustered by using the DBSCAN algorithm, and the parameters are set to cluster them into two categories: one is the background noise, and the other is the potential partial discharge point. The data points of the background noise are eliminated, and the potential partial discharge points are retained;

[0009] Step 4: Add a window, count the number of data points that meet the requirements in the sliding window along the phase, and screen the data points in each window based on the number of data points in the window and the amplitude dimension integration analysis;

[0010] Step 5, collecting and screening the images of the data points in each window, performing regular similarity analysis on the data points in each window and removing the interference points in the window by association, and finally obtaining the data points in the window of the real partial discharge point;

[0011] Step 6, adding the number of data points in the window for obtaining the real partial discharge point to obtain the final number of TEV pulses for output.

[0012] As a further description of the above technical solution: downsampling the collected TEV data to 5000 data points is as follows:

[0013] The collected TEV signal is processed with the data of 1s duration;

[0014] The data is reconstructed into a matrix form according to the period and phase, and then the original large matrix is ​​downsampled to a small matrix by taking the maximum value at equal intervals along the phase.

[0015] As a further description of the above technical solution: adding a window, the number of data points that meet the requirements in the phase sliding statistical window is specifically:

[0016] Add a window along the x-axis with a step size of 1, where the window length is less than or equal to 1 / 4 period;

[0017] Set the amplitude threshold of the data set, count the number of data points that meet the amplitude threshold window, and mark them as , , .... ,in Indicates the number of data points that meet the amplitude threshold in the nth window.

[0018] As a further description of the above technical solution: screening the data points in each window according to the number of data points in the window and the integrated analysis of the amplitude dimension specifically includes the following steps:

[0019] Calculate the threshold difference between the number of data points that meet the amplitude threshold and the total amount in each window ;

[0020] Calculate the mean difference of the stability coefficient of the amplitude of the data points in each window ;

[0021] Threshold Difference Difference from mean The accuracy coefficient of each data point in the window is calculated by weighted summation, and the data points in each window are screened based on the accuracy coefficient.

[0022] As a further description of the above technical solution: Calculate the threshold difference between the number of data points that meet the amplitude threshold and the total amount in each window Specifically:

[0023] Set the ratio threshold ;

[0024] Calculate the ratio of the number of data points that meet the amplitude threshold in each window to the total number : ;

[0025] Calculate the ratio of the number of data points that meet the amplitude threshold in each window to the total number and ratio threshold The difference , ,in is the default value, =0.3.

[0026] As a further description of the above technical solution: Calculate the mean difference of the stability coefficient of the amplitude of the data points in each window Specifically:

[0027] Calculate the amplitude data corresponding to the data points that meet the amplitude threshold in each window, and integrate to obtain the amplitude data set of the data points in each window, where the amplitude data set of the data point in the nth window is , ∈( ), represents the amplitude data of the mth data point in the nth window;

[0028] Based on the amplitude data set of each data point in the window, the stability coefficient of the amplitude of each data point in the window is calculated and marked as , , ... , and integrate to calculate the stability coefficient mean of the amplitude of each data point in the window ;

[0029] Calculate the difference between the stability coefficient of the data point amplitude in the window and the mean ,in, .

[0030] As a further description of the above technical solution: the threshold difference Difference from mean The weighted sum is performed to calculate the accuracy coefficient of each data point in the window, and the data points in each window are screened based on the accuracy coefficient:

[0031] Set the accuracy coefficient judgment threshold ;

[0032] Based on the difference threshold Difference from mean Calculate the accuracy coefficient of the corresponding data point in the window , accuracy coefficient The calculation formula is: ,That , is the weight factor, , All are greater than 0;

[0033] Compare and Size, sieving to obtain The corresponding data points in each window, and The corresponding data points in each window.

[0034] As a further description of the above technical solution: collecting and screening the images of the data points in each window, and performing regular similarity analysis and removal processing on the interference points in the data points in each window, and finally obtaining the data points in the window of the real partial discharge point specifically includes the following steps:

[0035] collection The corresponding data points in each window are integrated and analyzed to obtain a corrected image;

[0036] collection The corresponding data points in each window are corrected by The corresponding data points in each window are corrected and adjusted, and finally the data points in the window of the real partial discharge point are obtained.

[0037] As a further description of the above technical solution: Collection The corresponding data points in each window are integrated and analyzed to obtain the corrected image:

[0038] Retrieve The corresponding images of the data points in each window are plotted in the same coordinate system;

[0039] Collect the number of overlaps of data points in each window, collect data points with an overlap greater than 0.75X, and retain the collected data points in the coordinate system to obtain a corrected image, where X represents the number of overlaps that satisfy Number of windows.

[0040] As a further description of the above technical solution: Collection The corresponding data points in each window are corrected by The corresponding data points in each window are corrected and adjusted, and the data points in the window of the actual partial discharge point are finally obtained as follows: The corresponding data point images in each window are plotted on the corrected image, and the data points in the window that overlap with the data points on the corrected image are removed to obtain the data points in the window of the real partial discharge point.

[0041] In the above technical scheme, a TEV pulse sampling and counting method provided by the present invention has the following beneficial effects: the TEV pulse sampling and counting method is based on the TEV pulse counting method under undersampling conditions of the DBSCAN algorithm, which solves the problem of how to realize the statistics of the number of TEV pulses under undersampling conditions, and is of great significance to the miniaturization and practical application of equipment based on the TEV detection method. By calculating the threshold difference of the ratio of the number of data points that meet the amplitude threshold in each window to the total amount; and calculating the mean difference of the stability coefficient of the amplitude of the data points in each window, the accuracy coefficient of the data points in each window is calculated by weighted summing the threshold difference and the mean difference to screen the data points in each window, and realizing the correlation and integration analysis of the two dimensions of the number of data points in each window and the amplitude of each data point, the accuracy and comprehensiveness of the interference point removal are significantly improved, and the number of TEV pulses can be better counted. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0043] Figure 1 A schematic flow chart of a TEV pulse sampling and counting method provided in an embodiment of the present invention;

[0044] Figure 2 A PRPD diagram drawn for downsampled TEV data provided by an embodiment of the present invention;

[0045] Figure 3 A schematic diagram of the DBSCAN clustering results provided by an embodiment of the present invention;

[0046] Figure 4 A schematic diagram of a window for counting data points after removing background noise provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0048] See also Figure 1-Figure 4 The embodiment of the present invention provides a technical solution: a TEV pulse sampling and counting method, comprising the following steps:

[0049] Step 1: Collect TEV signal under undersampling condition, and set the sampling rate to 250K;

[0050] Step 2: Downsample the collected TEV data to 5000 data points; specifically: process the collected TEV signal with a duration of 1s; reconstruct the data into a matrix form according to the period and phase, and then downsample the original large matrix to a small matrix by taking the maximum value at equal intervals along the phase. Since the sampling rate is 250K, the original large matrix is ​​50*5000, the target matrix is ​​50*100, and the maximum value is taken at an interval of 100;

[0051] Step 3: If Figure 2 As shown in the PRPD diagram drawn based on TEV data, the data points in the PRPD diagram drawn based on TEV data are clustered using the DBSCAN algorithm, as shown in Figure 3 As shown, it should be noted that the well-known DBSCAN is a density-based spatial clustering algorithm, which divides areas with sufficiently high density into clusters. According to the characteristics of TEV data, appropriate eps and minPts parameters are set. EPS should be set to a neighborhood radius that can distinguish different TEV data, and minPts is determined according to the density distribution of TEV data. The parameters are set to cluster them into two categories: one is background noise, and the other is potential partial discharge points. The data points of the background noise are eliminated, and the potential partial discharge points are retained. Specifically, the data points in the PRPD map are clustered into two categories of background noise and potential partial discharge points. First, the PRPD map is observed to understand the distribution characteristics of the background noise and potential partial discharge points, and preliminary clustering is performed: a relatively large eps and a moderate MinPts are selected for preliminary clustering, and the clustering results are observed. Adjust the parameters: according to the results of the preliminary clustering, eps and MinPts are gradually adjusted until the background noise and potential partial discharge points can be clearly distinguished;

[0052] Step 4: Figure 4 As shown, windowing is performed, and the number of data points that meet the requirements in the sliding window is counted along the phase. Specifically, the number of data points that meet the requirements in the sliding window is as follows: windowing is performed along the x-axis with a step size of 1, and the window length is less than or equal to 1 / 4 cycle; the amplitude threshold of the data set is set, and the number of data points that meet the amplitude threshold window is counted, and they are marked as , , .... ,in Indicates the number of data points that meet the amplitude threshold in the nth window. The data points in each window are screened based on the number of data points in the window and the amplitude dimension integration analysis;

[0053] Step 5, collect and screen the images of the data points in each window, perform regular similarity analysis on the data points in each window, and associate and remove the interference points in the window, and finally obtain the data points in the window of the real partial discharge point. The interference points generated by the interference source will not have obvious phase concentration, and their distribution range is wide, covering the entire cycle. By performing regular similarity analysis on the data points in each window screened as interference points, the interference points in each window corresponding to the partial discharge point are then associated and removed;

[0054] Step 6, adding the number of data points in the window for obtaining the real partial discharge point to obtain the final number of TEV pulses for output.

[0055] The TEV pulse sampling and counting method is based on the TEV pulse counting method under under-sampling conditions of the DBSCAN algorithm, which solves the problem of how to realize the statistics of the number of TEV pulses under under-sampling conditions, and is of great significance to the miniaturization and practical application of equipment based on the TEV detection method.

[0056] Furthermore, the data points in each window are screened according to the number of data points in the window and the integrated analysis of the amplitude dimension, which specifically includes the following steps:

[0057] Calculate the threshold difference between the number of data points that meet the amplitude threshold and the total amount in each window ;

[0058] Specifically:

[0059] Set the ratio threshold ;

[0060] Calculate the ratio of the number of data points that meet the amplitude threshold in each window to the total number : ;

[0061] Calculate the ratio of the number of data points that meet the amplitude threshold in each window to the total number and ratio threshold The difference , ,in is the default value, =0.3.

[0062] Calculate the mean difference of the stability coefficient of the amplitude of the data points in each window , specifically:

[0063] Calculate the amplitude data corresponding to the data points that meet the amplitude threshold in each window, and integrate to obtain the amplitude data set of the data points in each window, where the amplitude data set of the data point in the nth window is , ∈( ), Represents the amplitude data of the mth data point in the nth window;

[0064] Based on the amplitude data set of each data point in the window, the stability coefficient of the amplitude of each data point in the window is calculated and marked as , , ... , and integrate to calculate the stability coefficient mean of the amplitude of each data point in the window ;

[0065] Calculate the difference between the stability coefficient of the data point amplitude in the window and the mean ,in, .

[0066] Threshold Difference Difference from mean The accuracy coefficient of each data point in the window is calculated by weighted summation, and the data points in each window are screened based on the accuracy coefficient, specifically:

[0067] Set the accuracy coefficient judgment threshold ;

[0068] Based on the difference threshold Difference from mean Calculate the accuracy coefficient of the corresponding data point in the window , accuracy coefficient The calculation formula is: ,That , is the weight factor, , are all greater than 0; it should be noted that the size of the weight coefficient is a specific value obtained by quantifying each data to facilitate subsequent comparison. The size of the weight coefficient depends on the number of comprehensive parameters and the technical personnel in this field preliminarily set the corresponding weight coefficient for each set of comprehensive parameters.

[0069] Compare and Size, sieving to obtain The corresponding data points in each window, and The corresponding data points in each window, The corresponding data points in each window are potential partial discharge points including interference points. The corresponding data points in each window are interference points.

[0070] The TEV pulse sampling and counting method calculates the threshold difference of the ratio of the number of data points that meet the amplitude threshold in each window to the total amount; and calculates the mean difference of the stability coefficient of the amplitude of the data points in each window. The accuracy coefficient of the data points in each window is calculated by weighted summing the threshold difference and the mean difference to screen the data points in each window, and realizes the correlation and integration analysis of the two dimensions of the number of data points in each window and the amplitude of each data point, which significantly improves the accuracy and comprehensiveness of interference point removal and can better count the number of TEV pulses.

[0071] Furthermore, the images of the data points in each window are collected and screened, and the interference points in the data points in each window are removed by regular similarity analysis, and finally the data points in the window of the real partial discharge point are obtained, which specifically includes the following steps:

[0072] collection The corresponding data points in each window are integrated and analyzed to obtain the corrected image, specifically:

[0073] Retrieve The corresponding images of the data points in each window are plotted in the same coordinate system;

[0074] Collect the number of overlaps of data points in each window, collect data points with an overlap greater than 0.75X, and retain the collected data points in the coordinate system to obtain a corrected image, where X represents the number of overlaps that satisfy Number of windows.

[0075] collection The corresponding data points in each window are corrected by The corresponding data points in each window are corrected and adjusted to finally obtain the data points in the window of the real partial discharge point. Specifically: The corresponding data point images in each window; the data points in each window are plotted on the corrected image, and the data points in the window that overlap with the data points on the corrected image are removed to obtain the data points in the window of the real partial discharge point.

[0076] The TEV pulse sampling and counting method collects The corresponding data points in each window are integrated and analyzed to obtain the corrected image. The corresponding data points in each window are corrected and adjusted, and finally the data points in the window of the real partial discharge point are obtained, and the interference points in the data points in each window are analyzed and removed based on the regular similarity, which further improves the accuracy of TEV pulse quantity statistics.

[0077] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A TEV pulse sampling and counting method, characterized in that: The following steps are involved: Step 1: Collect TEV signal under undersampling condition, and set the sampling rate to 250K; Step 2: Downsample the collected TEV data to 5000 data points; Step 3: Based on the PRPD diagram drawn by TEV data, the data points in the PRPD diagram drawn by TEV data are clustered by using the DBSCAN algorithm, and the parameters are set to cluster them into two categories: one is the background noise, and the other is the potential partial discharge point. The data points of the background noise are eliminated, and the potential partial discharge points are retained; Step 4: Add window, count the number of data points that meet the requirements in the sliding window along the phase, and screen the data points in each window according to the number of data points in the window and the amplitude dimension integration analysis. The specific number of data points that meet the requirements in the sliding window along the phase is: add window along the x-axis with a step size of 1, where the window length is less than or equal to 1 / 4 cycle; set the amplitude threshold of the data set, count the number of data points in the window that meet the amplitude threshold, and mark them as , , .... ,in Indicates the number of data points that meet the amplitude threshold in the nth window; Step 5, collecting and screening the images of the data points in each window, performing regularity similarity analysis on the data points in each window and associating and removing the interference points in the window, and finally obtaining the data points in the window of the real partial discharge point; The screening of the data points in each window according to the number of data points in the window and the integrated analysis of the amplitude dimension specifically includes the following steps: Calculate the threshold difference between the number of data points that meet the amplitude threshold and the total amount in each window ; Calculate the mean difference of the stability coefficient of the amplitude of the data points in each window ; Threshold Difference Difference from mean Perform weighted summation to calculate the accuracy coefficient of the data points in each window, and screen the data points in each window based on the accuracy coefficient; Calculate the threshold difference between the number of data points that meet the amplitude threshold and the total amount in each window Specifically: Set the ratio threshold ; Calculate the ratio of the number of data points that meet the amplitude threshold in each window to the total number : ; Calculate the ratio of the number of data points that meet the amplitude threshold in each window to the total number and ratio threshold The difference , ,in is the default value, =0.3; Calculate the mean difference of the stability coefficient of the amplitude of the data points in each window Specifically: Calculate the amplitude data corresponding to the data points that meet the amplitude threshold in each window, and integrate to obtain the amplitude data set of the data points in each window, where the amplitude data set of the data point in the nth window is , ∈( ), represents the amplitude data of the mth data point in the nth window; Based on the amplitude data set of each data point in the window, the stability coefficient of the amplitude of each data point in the window is calculated and marked as , , ... , and integrate to calculate the stability coefficient mean of the amplitude of each data point in the window ; Calculate the difference between the stability coefficient of the amplitude of the data point in the window and the mean ,in, ; Threshold Difference Difference from mean The weighted sum is performed to calculate the accuracy coefficient of each data point in the window, and the data points in each window are screened based on the accuracy coefficient: Set the accuracy coefficient judgment threshold ; Based on the difference threshold Difference from mean Calculate the accuracy coefficient of the corresponding data point in the window , accuracy coefficient The calculation formula is: ,That , is the weight factor, , All are greater than 0; Compare and Size, sieving to obtain The corresponding data points in each window, and The corresponding data points in each window; The images of the data points in each window are collected and screened, and the regular similarity analysis is performed on the data points in each window, and the interference points in the window are removed by association, so as to finally obtain the data points in the window of the real partial discharge point. Specifically, the following steps are included: collection The corresponding data points in each window are integrated and analyzed to obtain a corrected image; collection The corresponding data points in each window are corrected by The corresponding data points in each window are corrected and adjusted, and finally the data points in the window of the real partial discharge point are obtained; collection The corresponding data points in each window are integrated and analyzed to obtain the corrected image: Retrieve The corresponding data point images in each window are plotted in the same coordinate system; the overlap times of the data points in each window are collected, and the data points whose overlap times are greater than 0.75X are collected, and the collected data points are retained in the coordinate system to obtain a corrected image, where X represents the data points that meet Number of windows; collection The corresponding data points in each window are corrected by The corresponding data points in each window are corrected and adjusted, and the data points in the window of the actual partial discharge point are finally obtained as follows: The corresponding data point images in each window; the data points in each window are plotted on the corrected image, and the data points in the window that overlap with the data points on the corrected image are removed to obtain the data points in the window of the real partial discharge point; Step 6, adding the number of data points in the window for obtaining the real partial discharge point to obtain the final number of TEV pulses for output.

2. A TEV pulse sampling and counting method according to claim 1, characterized in that: The collected TEV data is downsampled to 5000 data points as follows: The collected TEV signal is processed with the data of 1s duration; The data is reconstructed into a matrix form according to the period and phase, and then the original large matrix is ​​downsampled to a small matrix by taking the maximum value at equal intervals along the phase.

Images