Partial discharge pulse signal extraction method

Abstract

The invention discloses a partial discharge pulse signal extraction method comprising the following steps: step 1, collecting a first signal which is an original pulse signal; step 2, performing combined balance noise reduction and filtering processing on the first signal to obtain a second signal which is a complete pulse signal; 3, a signal threshold value is set, a third signal is extracted from the second signal, and the third signal is an effective pulse signal; and step 4, SVD noise reduction is carried out on the third signal, and a fourth signal is output. According to the method, the signal threshold value is set, the effective partial discharge waveform is extracted, and the partial discharge signal data volume for training is remarkably reduced; the calculation speed of the device is accelerated, and the response rate of the device is improved; and the memory occupation of the collected data is reduced, so that the device can store more waveform data and effective information.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment condition monitoring, and particularly to a method for extracting partial discharge pulse signals. Background Art

[0002] Long-term partial discharge will damage the insulation structure, cause insulation aging, and ultimately develop and cause insulation failure. A large number of research results on GIS partial discharge detection show that in the high-pressure gas inside the equipment, the duration of the partial discharge pulse is very short, and the rise time of the wavefront is only about 1 ns. Such a very short-duration steep pulse causes great trouble to detection. Therefore, it is necessary to arrange an all-weather online monitoring system to capture such occasional pulse signals. However, due to the excessive amount of data, the acquisition and diagnosis efficiency is low.

[0003] However, the ultra-high frequency signals detected on-site are often affected by white noise, periodic interference, etc., resulting in difficult signal extraction. Moreover, the amount of data collected by the online monitoring system is huge, which will consume a large amount of system storage and communication resources, and poses high requirements on the sampling and storage capabilities of the monitoring system. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for extracting partial discharge pulse signals to solve the problems of low sampling and storage space and low diagnosis efficiency of the monitoring system.

[0005] The present invention adopts the following technical solutions to solve the technical problems:

[0006] A method for extracting partial discharge pulse signals includes: Step 1, collecting a first signal, where the first signal is an original pulse signal; Step 2, performing combined balanced noise reduction and filtering processing on the first signal to obtain a second signal, where the second signal is a complete pulse signal; Step 3, setting a signal threshold and extracting a third signal from the second signal, where the third signal is an effective pulse signal; Step 4, performing SVD noise reduction on the third signal and outputting a fourth signal.

[0007] Preferably, the threshold determination method is as follows: taking the threshold as the horizontal axis and the power ratio as the vertical axis, making a threshold power ratio curve graph, calculating the curve slope, and outputting the threshold; where the threshold is td and the slope is r.

[0008] Preferably, the power ratio calculation formula is: where the power ratio is W pr , W c is the power of the intercepted waveform, and W s is the power of the sampled waveform.

[0009] Preferably, the threshold output condition is that the curve slope becomes smaller and smaller and approaches a fixed value.

[0010] Preferably, the fixed value is taken as 0.3, and the output threshold is 0.36.

[0011] Preferably, the validity is set to evaluate the signal extraction effect, and the calculation formula is: V validity =(W v / W f )×100%, where the validity is V validity , W v is the power of the effectively extracted waveform, and W f is the power of the complete waveform.

[0012] Preferably, the calculated validity is 95.9.

[0013] By setting the signal threshold, the present invention extracts the effective partial discharge waveform, significantly reducing the amount of partial discharge signal data for training; accelerating the calculation speed of the device, improving the response rate of the device; reducing the memory occupancy of the collected data, enabling the device to store more waveform data and effective information. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a schematic flow chart of the method of the present invention;

[0015] Figure 2 is a curve graph of the threshold power ratio of the present invention;

[0016] Figure 3 is a schematic diagram of the extraction of the simulated partial discharge pulse signal in Embodiment 2 of the present invention;

[0017] Figure 4 is a schematic diagram of the extraction of the on-site partial discharge pulse signal in Embodiment 3 of the present invention;

[0018] Figure 5 is a schematic diagram of the original on-site partial discharge pulse signal in Embodiment 3 of the present invention;

[0019] Figure 6 is a schematic diagram of the effective on-site partial discharge pulse signal in Embodiment 3 of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0020] The technical solution of the present invention will be further described below in conjunction with the embodiments and the drawings.

[0021] A method for extracting partial discharge pulse signals includes: Step 1, collecting a first signal, where the first signal is an original pulse signal; Step 2, performing combined balanced noise reduction and filtering on the first signal to obtain a second signal, where the second signal is a complete pulse signal; Step 3, setting a signal threshold and extracting a third signal from the second signal, where the third signal is an effective pulse signal; Step 4, performing SVD noise reduction on the third signal and outputting a fourth signal.

[0022] When performing partial discharge detection on-site for power equipment, after collecting the original pulse signal, due to the relatively large amount of on-site environmental noise interference, joint balanced noise reduction and filtering processing are first carried out to obtain a complete pulse signal. At this time, if the complete pulse signal is directly used to generate a prpd pattern for signal diagnosis, it will result in a large amount of data for analysis and diagnosis, affecting the diagnosis efficiency and accuracy. Therefore, in the present invention, based on step 3, a signal threshold is set. Signal segments are extracted from each segment of the complete pulse signal to obtain a third signal, that is, an effective pulse signal. The effective pulse signal contains signal segments with partial discharge from the local discharge source and removes the noise parts at the head and tail of the signal. Therefore, it can greatly reduce the data size of a single partial discharge signal, enabling the monitoring system to store more data, and further analyze more partial discharge signals, greatly improving the efficiency of partial discharge signal diagnosis.

[0023] In a further implementation manner of this embodiment, the method for determining the threshold is as follows: Set the threshold as the horizontal axis and the power ratio as the vertical axis to create a threshold power ratio curve graph, calculate the curve slope, and output the threshold; where the threshold is td and the slope is r.

[0024] In a further implementation manner of this embodiment, the power ratio calculation formula is: Where the power ratio is W pr , W c is the power of the intercepted waveform, and W s is the power of the sampled waveform.

[0025] In a further implementation manner of this embodiment, the threshold output condition is that the curve slope becomes smaller and smaller and approaches a fixed value; the smaller the curve slope, the larger the threshold, and the more noise segments are cut off, while retaining the signal segments containing the partial discharge pulse signal. Therefore, when the curve slope approaches a fixed value, the threshold can be output.

[0026] In a further implementation manner of this embodiment, the fixed value ≤ 0.5. At this time, the output threshold is used to intercept the complete pulse signal, which can remove a large part of the noise and retain the core partial discharge pulse signal segment.

[0027] In a further implementation manner of this embodiment, since the on-site is inevitably filled with a large amount of noise and pure partial discharge signals cannot be detected, in order to evaluate the extraction effect of the true and effective pulse signal waveform, an effectiveness is set to evaluate the signal extraction effect. The calculation formula is: V validity =(W v / W f )×100%, where the effectiveness is V validity , W v is the power of the effectively extracted waveform, and W f is the power of the complete waveform.

[0028] Example 2

[0029] As Figure 3 shown, the signal schematic diagram of applying the simulated partial discharge pulse signal to the extraction method of the present invention is as follows. From top to bottom, they are the schematic diagrams of the analog signal, adding white noise, SVD noise reduction, S-G filter noise reduction, and joint balanced noise reduction of the pulse signal. The effective waveforms calculated by the joint balanced noise reduction and SVD noise reduction are respectively between the two red and two green dotted lines.

[0030] In this embodiment, the fixed value is taken as 0.3, and the output threshold is 0.36, which is used to extract the complete pulse signal. It can be known from Figure 3 that the accuracy of the effective pulse signal extracted after the joint balanced noise reduction is higher than that of the effective pulse signal only after the SVD noise reduction. The calculated effectiveness is 95.9%. Compared with the waveform extraction effectiveness of 80% after relying solely on the SVD noise reduction, it has increased by 15.9 percentage points, and the extraction effect is remarkable.

[0031] Example 3,

[0032] As Figure 4 shown, the signal schematic diagram of applying the partial discharge pulse signal collected on-site to the extraction method of the present invention is as follows. From top to bottom, they are the schematic diagrams of the on-site signal, adding white noise, SVD noise reduction, S-G filter noise reduction, and joint balanced noise reduction of the pulse signal;

[0033] In this embodiment, the fixed value is taken as 0.3, and the output threshold td is 0.05, which is used to extract the complete pulse signal. It can be known from Figure 4 that the accuracy of the effective pulse signal extracted after the joint balanced noise reduction is higher than that of the effective pulse signal only after the SVD noise reduction. The calculated effectiveness is 91.23%. Considering the influence of noise, the relative effectiveness of 91.23% is a reasonable effective extraction result. The results of identifying and locating the partial discharge defects using the extracted effective pulses are consistent with the complete signal. The extraction method of the present invention is feasible and effective, as shown in Chart 1;

[0034] Table 1 Diagnostic results of the complete pulse signal and the effective pulse signal

[0035]

[0036] As Figure 5 and Figure 6 shown, they are the input and output waveform diagrams of the on-site partial discharge pulse signal. The size of the input waveform is 129k, and the size of the output waveform is 35.8k. The size of the effective pulse signal is only 27% of the input waveform, which greatly reduces the storage requirements for the monitoring system. Under the same storage space conditions, more effective pulse signals can be diagnosed, improving the diagnostic efficiency.

[0037] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for extracting a partial discharge pulse signal, characterized in that: include: Step 1, collecting a first signal, where the first signal is an original pulse signal; Step 2, perform joint balanced noise reduction and filtering on the first signal to obtain a second signal, where the second signal is a complete pulse signal; Step 3, set a signal threshold to extract a third signal from the second signal, where the third signal is a valid pulse signal; Step 4, perform SVD noise reduction on the third signal to output a fourth signal.

2. The method for extracting partial discharge pulse signals according to claim 1, characterized in that: The threshold determination method is to set the threshold as the horizontal axis and the power ratio as the vertical axis, draw a threshold power ratio curve, calculate the slope of the curve, and output the threshold; wherein the threshold is td and the slope is r.

3. The method for extracting partial discharge pulse signals according to claim 1, characterized in that: The power ratio calculation formula is: Among them, the power ratio is W pr , W c is the intercepted waveform power, W s is the sampled waveform power.

4. The method for extracting partial discharge pulse signals according to claim 2, characterized in that: The threshold output condition is that the slope of the curve becomes smaller and smaller and approaches a fixed value.

5. The method for extracting partial discharge pulse signals according to claim 4, characterized in that: The fixed value is ≤0.

5.

6. The method for extracting partial discharge pulse signals according to claim 1, characterized in that: Set the validity to evaluate the signal extraction effect. The calculation formula is: V validity =(W v / W f )×100%, where the effective degree is V validity , W v To effectively extract the waveform power, W f is the complete waveform power.

7. The method for extracting partial discharge pulse signals according to claim 6, characterized in that: The calculated validity is 95.

9.

8. The method for extracting partial discharge pulse signals according to claim 5, characterized in that: The fixed value is 0.3, td=0.36.

Images