A Partial Discharge Pulse Burst Abnormal Waveform Identification Method Based on Trigger Threshold Moving Window

Abstract

This invention relates to a method for identifying abnormal waveforms of partial discharge pulse groups based on a trigger threshold moving window, comprising: Step 1, calculating the average amplitude of a points at the beginning and b points at the end of the recorded waveform to obtain the reference offset value of the recorded waveform; Step 2, removing the reference offset value from the recorded waveform to form a processed waveform; Step 3, establishing an adjustable-duration trigger threshold moving window based on the trigger threshold Yz, and calculating the time standard deviation T and frequency standard deviation B of the data within the moving window when the absolute amplitude within the moving window is greater than or equal to Yz / m, and recording the extreme values ​​and corresponding positions when T×B≥1; Step 4, when the number of extreme values ​​k... j When the value is greater than 1, the time-domain waveform of the currently recorded single pulse is judged as an abnormal waveform; when the number of extreme values ​​k j When the value equals 1, the time-domain waveform of the currently recorded single pulse is determined to be a normal waveform. Compared with the prior art, the present invention has advantages such as good robustness.

Description

Technical Field

[0001] This invention relates to a method for identifying abnormal waveforms of partial discharge pulse bursts, and more particularly to a method for identifying abnormal waveforms of partial discharge pulse bursts based on a trigger threshold moving window. Background Technology

[0002] Partial discharge (PD) pulse source detection technology based on ultra-wideband (the bandwidth can cover or involve the frequency bands defined by GB / T 7354, GB / T 20833.1, GB / T 20833.2 and GB / T 23642 for conventional electrical testing (30kHz-1MHz), high frequency (3MHz-30MHz) and very high frequency (30MHz-300MHz)) is now widely used for offline or online testing of electrical equipment such as generators, transformers, cables and surge arresters.

[0003] This technology replaces traditional pulse peak-time series detection with pulse waveform-time series detection, recording the waveform of a single pulse and its acquisition time (phase information can be used under AC voltage). Because ultra-wideband detection retains more complete waveform information, and waveforms from different pulse sources exhibit self-similarity, a certain method can be used to quickly classify the acquired mixed raw pulse groups, enabling the separation of pulse sources, i.e., the separation of PD sources from noise sources or multiple PD sources. The State Grid Corporation of China's enterprise standard, "Q / GDW11400-2015 Field Application Guidelines for High-Frequency Partial Discharge On-Cellular Testing Technology for Power Equipment," provides the following in Appendix B: Figure 1 The typical waveforms of a single pulse corresponding to the corona discharge, internal discharge, and surface discharge spectrum characteristics shown are presented (sampling rate 100 MS / s, analog bandwidth 30 MHz). Based on the time-domain waveform of each pulse in the pulse waveform-time series, feature parameters of the pulse group are extracted, and cluster analysis is performed on the corresponding feature parameters in the 2D plane or 3D space. This separates the pulse group into sub-pulse groups with their own characteristics, thereby achieving the separation of multiple PD sources and noise sources, laying the foundation for further research on the identification of PD types and severity within power equipment.

[0004] However, in practical applications, the waveform generated by the PD source, within the 1μs or 2μs acquisition time set on the detection instrument, will exhibit distorted multi-peak single pulse waveforms, pulse superposition, or even multiple consecutive pulses, etc. Figure 1The single exponential oscillating pulse waveform shown represents anomalous pulse waveforms with significant time-domain differences. These anomalous pulse waveforms severely hinder the use of current methods for extracting pulse waveform feature parameters based on equivalent frequency and equivalent duration, thus affecting subsequent clustering analyses based on feature parameters in 2D or 3D space. This prevents ultra-wideband pulse source detection techniques from achieving complete separation of multiple PD sources and noise sources. However, PD sources are random, and anomalous pulse waveforms are unavoidable during pulse group detection and recording. Therefore, how to identify anomalous waveforms within a pulse group in real-time and rapidly is a pressing problem that needs to be solved. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art by providing a method for identifying abnormal waveforms of partial discharge pulse groups based on a trigger threshold moving window.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] According to one aspect of the present invention, a method for identifying anomalous waveforms of partial discharge pulse bursts based on a trigger threshold moving window is provided. This method is used for recording the time-domain waveform of a single pulse in ultra-wideband detection, employing a threshold triggering method and using sampling points within the first 20% and last 80% of the time period before the triggering time to form the recorded waveform p. j (t i The identification method includes the following steps:

[0008] Step 1: For the a points at the beginning of the recorded waveform and the b points at the end of the recorded waveform, calculate the average amplitude and then obtain the reference offset value Pz of the recorded waveform.

[0009] Step 2, record waveform p j (t i Remove the reference bias value from the waveform to form the processed waveform p' j (t i );

[0010] Step 3: Based on the trigger threshold Yz, establish an adjustable duration trigger threshold moving window. For the processed waveform from the beginning to the end of the recording, when there is an absolute amplitude greater than or equal to Yz / m within the trigger threshold moving window, calculate the time standard deviation T and frequency standard deviation B of the data within the moving window. When T×B≥1, record the extreme value and the corresponding position.

[0011] Step 4, when the number of extreme values ​​k j When the value is greater than 1, the time-domain waveform of the currently recorded single pulse is judged as an abnormal waveform, marked, and displayed as a time-frequency waveform; when the number of extreme values ​​k j When the value is equal to 1, the time-domain waveform of the currently recorded single pulse is determined to be a normal waveform.

[0012] As a preferred technical solution, m = 1.25.

[0013] According to another aspect of the present invention, a system for the partial discharge pulse burst abnormal waveform identification method based on the trigger threshold moving window is provided, comprising an ultra-wideband detection and acquisition pulse waveform-time series module, a front reference module, a rear reference module, a trigger threshold moving window module, a marking module, and a discrimination and display module;

[0014] The ultra-wideband detection and acquisition pulse waveform-time series module is connected to the trigger threshold moving window module through the front reference module, the rear reference module, and the trigger threshold moving window module, the marking module, and the discrimination and display module are connected in sequence.

[0015] As a preferred technical solution, the ultra-wideband detection and acquisition pulse waveform-time series module employs a data acquisition device with an analog bandwidth of tens of MHz and a sampling rate of 100 MS / s or higher. Based on pulse waveform threshold triggering technology, it records the waveform of a single pulse in the time domain and the pulse waveform-time series corresponding to the trigger time, i.e., the pulse group p. j (t i ).

[0016] As a preferred technical solution, the p j (t i The definition is as follows:

[0017]

[0018] In the formula:

[0019] j is the j-th pulse, where j = 1, 2, ..., N, and N is the total number of pulse waveforms in the pulse group;

[0020] A i To record the amplitude at the i-th point in the waveform;

[0021] k represents a pulse waveform composed of k points, the number of which is determined by the sampling rate f. s *Sampling duration is determined.

[0022] As a preferred technical solution, the front reference module and the rear reference module record waveform p. j (t i Calculate the average amplitude of the first 'a' points and the last 'b' points to obtain the reference offset value Pz of the recorded waveform. The specific calculation is as follows:

[0023]

[0024] Equation (2) uses the beginning and end parts to calculate the average of the two amplitudes, and then sums them to obtain the average value.

[0025] As a preferred technical solution, the trigger threshold moving window module establishes an adjustable-duration trigger threshold moving window based on the trigger threshold Yz. For the processed waveform from the beginning to the end of the recording, when there is an absolute amplitude greater than or equal to Yz / m within the trigger threshold moving window, the time standard deviation T and frequency standard deviation B of the data within the moving window are calculated. When T×B≥1, the extreme value and the corresponding position are recorded.

[0026] As a preferred technical solution, the specific working process of the trigger threshold moving window module (13) is as follows:

[0027] 1) Set the duration to 1 / n of the waveform duration and the number of data points to k;

[0028] 2) The processed waveform is iteratively judged from the beginning to the end of the recording. When an absolute amplitude greater than or equal to Yz / m exists within the trigger threshold moving window, the signal y(t) within the moving window is calculated. i The time standard deviation T and frequency standard deviation B, where i = 1,...,k, are defined as follows:

[0029]

[0030]

[0031] in, For the signal y(t) within the moving window i ) energy;

[0032] For the signal y(t) within the moving window i (Time mean)

[0033] For the signal y(t) within the moving window i The frequency mean of )

[0034] Y(f i ) is y(t i DFT of )

[0035] If T×B≥1, then retain the trigger threshold moving window;

[0036] 4) If two adjacent trigger threshold moving windows are retained, only the one with the larger absolute amplitude is retained, thus completing the elimination of duplicate extreme values.

[0037] As a preferred technical solution, the marking module performs the above processing on all recorded waveforms in the pulse waveform-time series, i.e., the pulse group, and records each recorded waveform p.j (t i Extreme values ​​existing in ) and corresponding position

[0038] As a preferred technical solution, the discrimination and display module is used to record waveform p in a pulse group. j (t i Extreme values The number of k j When the value is greater than 1, the time-domain waveform of the currently recorded single pulse is judged as an abnormal waveform and marked; when the extreme value is... The number of k j When the value is equal to 1, the time-domain waveform of the currently recorded single pulse is determined to be a normal waveform.

[0039] Compared with the prior art, the present invention has the following advantages:

[0040] 1. This invention solves the problem that the current signal processing flow based on ultra-wideband PD pulse source detection technology does not yet consider the impact of distorted multi-peak single pulse waveforms, pulse superposition, or even multiple continuous abnormal pulse waveforms on the algorithm.

[0041] 2. The method of the present invention has the advantages of good robustness and simple algorithm that can be easily implemented.

[0042] 3. The trigger threshold moving window established in this invention ensures the accuracy of pulse extreme value detection by jointly judging the threshold and the time-frequency characteristics of the data within the moving window. Attached Figure Description

[0043] Figure 1 Examples of results provided in the appendix of the current standard;

[0044] Figure 2 These are the main components of the system shown in this invention;

[0045] Figure 3 This is a flowchart of pulse data processing in the method of the present invention;

[0046] Figure 4 This is a schematic diagram illustrating an example of the time-domain waveform processing result of a single pulse contained in a pulse group, according to the present invention.

[0047] Figure 5 Typical normal waveforms of oil-filled spike plate defect discharge and typical abnormal waveforms detected by the method of this invention are obtained for DC withstand voltage ultra-wideband detection.

[0048] Figure 6 Typical normal waveforms of internal air gap defect discharge in oil-impregnated paperboard and typical abnormal waveforms detected by the method of this invention are obtained for DC withstand voltage ultra-wideband detection.

[0049] Figure 7 Typical normal waveforms of surface air gap defect discharge in oil paper are obtained for DC withstand voltage ultra-wideband detection, as well as typical abnormal waveforms detected by the method of this invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0051] This invention provides a rapid real-time identification method for abnormal waveforms in partial discharge pulse bursts using a trigger threshold moving window. It enables the marking of abnormal pulse waveforms within pulse bursts recorded using ultra-wideband PD pulse source detection technology, facilitating their removal. For recording the time-domain waveform of a single pulse using ultra-wideband detection (typically 100 MS / s and above), a threshold Yz triggering method is employed, with the recorded waveform p composed of sampling points within the first 20% and last 80% of the pulse duration before and after the triggering time. j (t i First, for the recorded waveform at points a at the beginning and points b at the end, calculate the average amplitude to obtain the reference offset value Pz of the recorded waveform; second, from the recorded waveform p... j (t i Remove the reference bias value from the waveform to form the processed waveform p' j (t i ); Again, based on the trigger threshold Yz, an adjustable-duration trigger threshold moving window is established. For the processed waveform from the beginning to the end of the recording, when there is an absolute amplitude greater than or equal to Yz / m within the trigger threshold moving window (generally m = 1.25, i.e., 80% of the threshold), the time standard deviation T and frequency standard deviation B of the data within the moving window are calculated. When T × B ≥ 1, the extreme values ​​and corresponding positions are recorded; finally, when the number of extreme values ​​k j When the value is greater than 1, the time-domain waveform of the currently recorded single pulse is judged as an abnormal waveform, marked, and displayed in the time-frequency domain (manual verification); when the number of extreme values ​​k j When the value is equal to 1, the time-domain waveform of the currently recorded single pulse is determined to be a normal waveform.

[0052] like Figure 2 As shown, the system of the present invention for the partial discharge pulse group abnormal waveform identification method based on trigger threshold moving window includes an ultra-wideband detection and acquisition pulse waveform-time series module 10, a front reference module 11, a rear reference module 12, a trigger threshold moving window module 13, a marking module 14, and a discrimination and display module 15.

[0053] The ultra-wideband detection and acquisition pulse waveform-time series module 10 is connected to the trigger threshold moving window module 13 through the front reference module 11 and the rear reference module 12, respectively. The trigger threshold moving window module 13, the marking module 14 and the discrimination and display module 15 are connected in sequence.

[0054] The ultra-wideband detection and acquisition pulse waveform-time series module 10 is a data acquisition device using an analog bandwidth of tens of MHz and a sampling rate of 100 MS / s or higher. Based on pulse waveform threshold triggering technology, it records the waveform of a single pulse in the time domain and the pulse waveform-time series corresponding to the trigger time, i.e., the pulse group p. j (t i ), defined as follows:

[0055]

[0056] In the formula:

[0057] j — the j-th pulse (j = 1, 2, ..., N, where N is the total number of pulse waveforms in the pulse group);

[0058] A i —Record the amplitude (mV or pC) at the i-th point in the waveform;

[0059] k — The pulse waveform consists of k points, the number of points being determined by the sampling rate f. s *Sampling duration is determined.

[0060] Among them, the trigger threshold Yz is set in real time based on the background noise; it can also be determined according to the specified requirements of the power equipment PD test. Generally, the trigger threshold is 30% of the required value to avoid missing PD pulse signals. In addition, p j (t i The recorded waveform consists of sampling points within the 20% before and 80% after the trigger time. Figure 4 (a) is an example of a pulse waveform-time series, i.e., a single pulse time-domain waveform contained in a pulse group, obtained by a sampling rate of 250MS / s and an analog bandwidth of 50MHz. The trigger threshold is set to -5mV, and the duration of a single pulse time-domain waveform is 2kns, i.e. 2μs.

[0061] The aforementioned front reference module 11 and rear reference module 12 record waveform p j (t i Calculate the average amplitude of the first 'a' points and the last 'b' points, and obtain the reference offset value Pz of the recorded waveform, as follows:

[0062]

[0063] Equation (2) uses the average of the two amplitudes obtained from the beginning and end parts, and then sums them to obtain the average value, which has strong robustness. Set a = b = 10 points. Figure 4 (a) Record waveform p j (t i The processed waveform p' is formed by removing the reference bias value Pz from the input waveform. j (t i )like Figure 4 As shown in (b).

[0064] The trigger threshold moving window module 13 establishes an adjustable-duration trigger threshold moving window based on the trigger threshold Yz. For the processed waveform, from the beginning to the end of the recording, when an absolute amplitude greater than or equal to Yz / m exists within the trigger threshold moving window (generally m = 1.25, i.e., 80% of the threshold), the time standard deviation T and frequency standard deviation B of the data within the moving window are calculated. When T × B ≥ 1, the extreme values ​​and corresponding positions are recorded, as detailed below:

[0065] 1) Set the duration to 1 / n of the waveform duration and the number of data points to k;

[0066] 2) The processed waveform is iteratively judged from the beginning to the end of the recording. When an absolute amplitude greater than or equal to Yz / m exists within the trigger threshold moving window, the signal y(t) within the moving window is calculated. i The time standard deviation T and frequency standard deviation B, i = 1, ..., k, are defined as follows:

[0067]

[0068]

[0069] in, ——Moving the signal y(t) within the window i ) energy;

[0070] ——Moving the signal y(t) within the window i (Time mean)

[0071] ——Moving the signal y(t) within the window i The frequency mean of Y(f) i ) is y(t i The DFT of ).

[0072] If T×B≥1, then the trigger threshold moving window is retained.

[0073] 3) If two adjacent trigger threshold moving windows are retained, only the one with the larger absolute amplitude is retained, thus completing the elimination of duplicate extreme values.

[0074] Figure 4 (c) shows the following according to Figure 4 (b) shows the processed waveform p' j (t i The setup time is 200ns, which is 1 / 10 of the processed waveform p'. j (t i Example of detecting and retaining the extreme values ​​of the following two trigger threshold moving windows: duration and -4mV threshold.

[0075] The marking module 14 performs the above processing on all recorded waveforms in the pulse waveform-time series, i.e., the pulse group, and records each recorded waveform p. j (t i Extreme values ​​existing in ) and corresponding position

[0076] The discrimination and display module 15 records waveform p in the pulse group. j (t i Extreme values The number of k j When the value is greater than 1, the time-domain waveform of the currently recorded single pulse is judged as an abnormal waveform and marked; when the extreme value is... The number of k j When the value equals 1, the time-domain waveform of the currently recorded single pulse is determined to be a normal waveform; and the time-domain waveform and corresponding spectrum of the marked abnormal waveforms are displayed, allowing for further manual confirmation of the detection results.

[0077] Figures 5 to 7 The figures show typical normal waveforms obtained from detecting internal air gaps and surface defects of oil-impregnated paperboard in ultra-wideband DC withstand voltage tests at a sampling rate of 250 MS / s and a 50 MHz analog bandwidth, as well as typical abnormal waveforms detected by the method of this invention. Specifically, the pulse group corresponding to the negative polarity DC withstand voltage oil-impregnated paperboard defect test has a detection threshold of -10 mV, a sampling duration of 2 μs (2000 ns), a trigger threshold, and a moving window threshold of -8 mV. The detected abnormal waveforms are shown below. Figure 5 As shown in (b) to (h); the pulse group corresponding to the internal air gap defect test of negative polarity DC withstand voltage oil-impregnated paperboard, the detection threshold is -50mV, the sampling time is 2μs, the trigger threshold moving window threshold is -40mV, and the detected abnormal waveform is as follows. Figure 6 As shown in (b) to (h); the pulse group corresponding to the negative polarity DC withstand voltage oil paper surface defect test, the detection threshold is -5mV, the sampling time is 2μs, the trigger threshold moving window threshold is -4mV, and the detected abnormal waveform is as follows. Figure 7 As shown in (b) to (h).

[0078] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for identifying anomalous waveforms of partial discharge pulse bursts based on a trigger threshold moving window. This method is designed for recording the time-domain waveform of a single pulse in an ultra-wideband detection system. It employs a threshold triggering method, and the recorded waveform is composed of sampling points within the first 20% and last 80% of the pulse duration before the trigger moment. Its characteristics are, The identification method includes the following steps: Step 1, record the beginning portion of the waveform. Points and the ending part At each point, the average amplitude is calculated to obtain the reference offset value for recording the waveform. ; Step 2, from the recorded waveform The reference bias value is removed to form the processed waveform. ; Step 3, based on the trigger threshold An adjustable-duration trigger threshold moving window is established. For the processed waveform, from the beginning to the end of the recording, if an absolute amplitude greater than or equal to [a certain value] exists within the trigger threshold moving window... Calculate the time standard deviation of the data within the moving window. and frequency standard deviation ,when Record the extreme values ​​and their corresponding positions in real time; Step 4, when the number of extreme values When the value is greater than 1, the time-domain waveform of the currently recorded single pulse is judged as an abnormal waveform, marked, and displayed as a time-frequency waveform; when the number of extreme values ​​is... When the value equals 1, the time-domain waveform of the currently recorded single pulse is determined to be a normal waveform; The aforementioned The definition is as follows: （1） In the formula: For the first pulses, of which , This represents the total number of pulse waveforms contained in the pulse group. To record the first waveform The amplitude at each point; The pulse waveform is composed of It consists of 10 points, the number of points is determined by the sampling rate. *Sampling duration is determined; The front reference module (11) and the rear reference module (12) record the waveform. The beginning part Points and the ending part At each point, calculate the average amplitude to obtain the reference offset value for recording the waveform. The specific calculations are as follows: （2） Equation (2) uses the beginning and end parts to calculate the average of the two amplitudes, and then sums them to obtain the average value; Step 3 specifically includes: 1) Set the duration to the duration of the processed waveform. The number of data points is ; 2) The waveform is processed and checked cyclically from the beginning to the end of the recording. If an absolute amplitude greater than or equal to the threshold is found within the moving window, the waveform is checked. At that time, obtain the signal within the moving window. Time standard deviation and frequency standard deviation ,in The definition is as follows: （3） （4） in, For moving window signals Energy; For moving window signals The time mean; For moving window signals The frequency mean; for DFT; if If so, the trigger threshold moving window will be retained; 4) If two adjacent trigger threshold moving windows are retained, only the one with the larger absolute amplitude is retained, thus completing the elimination of duplicate extreme values.

2. The method for identifying abnormal waveforms of partial discharge pulse groups based on a trigger threshold moving window according to claim 1, characterized in that, The aforementioned .

3. A system for identifying abnormal waveforms of partial discharge pulse groups based on a trigger threshold moving window as described in claim 1, characterized in that, It includes an ultra-wideband pulse waveform-time series module (10), a front reference module (11), a rear reference module (12), a trigger threshold moving window module (13), a marking module (14), and a discrimination and display module (15). The ultra-wideband detection and acquisition pulse waveform-time sequence module (10) is connected to the trigger threshold moving window module (13) through the front reference module (11) and the rear reference module (12), respectively. The trigger threshold moving window module (13), the marking module (14) and the discrimination and display module (15) are connected in sequence.

4. The system according to claim 3, characterized in that, The ultra-wideband detection and acquisition pulse waveform-time series module (10) is a data acquisition device with an analog bandwidth of tens of MHz and a sampling rate of 100 MS / s or higher. Based on the pulse waveform threshold triggering technology, it records the waveform of a single pulse in the time domain and the pulse waveform-time series corresponding to the trigger time, i.e., the pulse group. .

5. The system according to claim 3, characterized in that, The trigger threshold moving window module (13) determines the trigger threshold based on the trigger threshold. An adjustable-duration trigger threshold moving window is established. For the processed waveform, from the beginning to the end of the recording, if an absolute amplitude greater than or equal to [a certain value] exists within the trigger threshold moving window... Calculate the time standard deviation of the data within the moving window. and frequency standard deviation ,when Record the extreme values ​​and their corresponding positions.

6. The system according to claim 3, characterized in that, The marking module (14) performs the above processing on all recorded waveforms in the pulse waveform-time series, i.e., the pulse group, and records each recorded waveform. Extreme values ​​existing in and corresponding position .

7. The system according to claim 3, characterized in that, The discrimination and display module (15) is used to record waveforms in a pulse group. Mid-extreme Number of If the value is greater than 1, the time-domain waveform of the currently recorded single pulse is identified as an abnormal waveform and marked accordingly; When extreme values Number of When the value is equal to 1, the time-domain waveform of the currently recorded single pulse is determined to be a normal waveform.

Images