Insulation partial discharge signal extraction method under high-frequency pulse voltage

Abstract

The invention discloses an insulation partial discharge signal extraction method under high-frequency pulse voltage, and aims to solve the problem of interference superposition of high-frequency pulse voltage partial discharge and edge overshoot thereof in engineering practice. Building a partial discharge test platform under high-frequency pulse voltage, and collecting partial discharge signals through an ultrahigh-frequency antenna; secondly, reconstructing white noise according to energy distribution of the collected signals, determining the number of wavelet decomposition layers according to the energy ratio of the pure signals to the reconstructed white noise, and meanwhile, designing a multi-scale threshold function to remove signal noise; and finally, taking a signal from voltage zero crossing to voltage stability as a potential aliasing signal, judging whether edge overshoot aliasing exists in the signal by using a cross correlation coefficient, if not, directly eliminating the power supply interference, and otherwise, designing a high-pass filter according to a pulse edge to eliminate the edge overshoot interference. By means of the method, interference components in the insulation partial discharge signals under the high-frequency impulse voltage can be effectively removed, and data support is provided for further analysis of the insulation performance.

Description

Technical Field

[0001] The invention belongs to the field of online insulation monitoring and fault diagnosis of power equipment, and relates to a method for extracting insulation partial discharge signals under high-frequency pulse voltage. Technical Background

[0002] Insulation degradation of electrical equipment is an important cause of equipment failure. Insulation defects come in various forms, and the degree of harm varies depending on the defect factors. In the actual operation of the power system, the insulation of the variable frequency motor to the ground is subjected to the influence of square wave voltage for a long time, and the electrical equipment is also subjected to the impact of high-frequency pulse overvoltage due to the frequent operation of switching devices such as circuit breakers and power electronic devices. Partial discharge testing is an effective means to evaluate and diagnose the insulation degradation state of electrical equipment. During the detection process, the rapid rise and fall of the square wave voltage will generate overcharge current, which may be mixed with the partial discharge signal and affect the test results. In order to solve the problem of superposition of high-frequency pulse voltage partial discharge and its edge overshoot interference, it is necessary to carry out research on relevant test platforms and explore discharge signal extraction technology. Summary of the invention

[0003] The present invention produces three types of insulation defects based on engineering practice and carries out partial discharge detection tests; builds a partial discharge test platform under high-frequency pulses, and collects partial discharge signals through ultra-high frequency antennas; reconstructs white noise according to the energy distribution of the collected signal, determines the number of decomposition layers according to the energy ratio of the collected signal to the reconstructed white noise, and designs a multi-scale threshold function to remove the signal white noise; takes the signal between voltage zero crossing and stability as a potential aliasing signal, and uses the mutual correlation coefficient to determine whether the signal has edge overshoot aliasing; if the information is not aliased, the power supply interference is directly removed, otherwise a high-pass filter is designed according to the pulse edge to eliminate the edge overshoot interference.

[0004] The present invention provides a method for extracting insulation partial discharge signals under high-frequency pulse voltage, the method comprising:

[0005] Step 1: Build a partial discharge test platform under high-frequency pulses and collect partial discharge signals through an ultra-high frequency antenna;

[0006] Step 2: Perform wavelet decomposition on the collected signal and reconstruct the white noise according to the energy distribution of the decomposed signal;

[0007] Step 3: Determine the decomposition level of the collected signal by the energy ratio of the pure signal and the reconstructed white noise at each wavelet scale;

[0008] Step 4: Design a multi-scale threshold function to remove signal noise in layers;

[0009] Step 5: Take the signal between voltage zero crossing and stability as the potential aliasing signal, and use the mutual correlation coefficient to determine whether the signal has edge overshoot aliasing. If the information is not aliased, directly remove the power supply interference. Otherwise, design a high-pass filter based on the pulse edge to eliminate the edge overshoot interference.

[0010] Optionally, in step 2, the collected partial discharge signal is decomposed by Dobesie No. 2 db2 wavelet, the energy of each layer of detail components is calculated by the following formula, the number of layers L of the detail component with the minimum energy is determined, and the white noise is reconstructed using all detail components with a number of layers less than or equal to L:

[0011]

[0012] Among them, X j [i] is the j-th layer detail component of the original discharge signal; Function Returns The j corresponding to the maximum value.

[0013] Optionally, in step 3, the number of decomposition layers is determined according to the energy ratio of the pure signal and the reconstructed white noise in each high-frequency signal layer:

[0014]

[0015] Among them, R is the number of wavelet decomposition layers; X j [i] is the j-th layer detail component of the original discharge signal; N j [l] is the j-th layer detail component of the reconstructed white noise.

[0016] Optionally, in step 4, the partial discharge signal X[n] is subjected to R-layer wavelet decomposition, and a multi-scale threshold function is designed to perform threshold processing on the wavelet coefficients of each layer. The multi-scale threshold function is as follows:

[0017]

[0018] Among them, T0 is the adaptive threshold, and the calculation formula is: X j [i] is the detail component of the jth layer of the partial discharge signal; |X j [i]| med is the median value of the detail component; L(X j [i]) is the length of the detail component.

[0019] Optionally, in step 5, a program is designed to determine whether the power supply interference information is aliased with the partial discharge information; N sampling point data between voltage zero crossing and stable state are taken as potential aliased signal s n, taking the first s1 as the reference signal, calculate the correlation coefficients of all potential aliasing signals; if the correlation coefficients are all greater than 0.95, it can be considered that the potential aliasing signals are all power supply interference at this time, and can be set to zero; if the correlation coefficient is less than or equal to 0.95 and the oscillation is severe, it is considered that the power supply interference signal is aliased with the partial discharge signal at this time, and the collected signal is filtered through a high-pass filter; reduce the pulse voltage to below the discharge voltage, and take the cutoff frequency of the test signal with a voltage rise time of 1μs as the reference, and design the cutoff frequency of the high-pass filter under different voltage rise times by the following formula:

[0020]

[0021] Among them, f cb is the cut-off frequency of the test signal when the voltage rise time is 1μs; T a is the voltage rise time. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flow chart of a method for extracting insulation partial discharge signals under high-frequency pulse voltage provided by an embodiment of the present invention;

[0023] Figure 2 A schematic diagram of a partial discharge test platform provided by an embodiment of the present invention;

[0024] Figure 3 A partial discharge detection signal diagram provided by an embodiment of the present invention;

[0025] Figure 4 This is a partial discharge denoising signal diagram provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] Figure 1 The flowchart of the method for extracting insulation partial discharge signals under high-frequency pulse voltage provided by the embodiment of the present invention is as follows: Figure 1 As shown, the method includes:

[0028] Step 1: Build a partial discharge test platform under high-frequency pulses and collect partial discharge signals through an ultra-high frequency antenna;

[0029] Step 2: Perform wavelet decomposition on the collected signal and reconstruct the white noise according to the energy distribution of the decomposed signal;

[0030] Step 3: Determine the decomposition level of the collected signal by the energy ratio of the pure signal and the reconstructed white noise at each wavelet scale;

[0031] Step 4: Design a multi-scale threshold function to remove signal noise in layers;

[0032] Step 5: Take the signal between voltage zero crossing and stability as the potential aliasing signal, and use the mutual correlation coefficient to determine whether the signal has edge overshoot aliasing. If the information is not aliased, directly remove the power supply interference. Otherwise, design a high-pass filter based on the pulse edge to eliminate the edge overshoot interference.

[0033] In the embodiment of the present invention, in step 1, a cable defect model is prepared and constructed as follows Figure 2 The test platform shown in the figure. In the system, the ZMC-100 high-frequency pulse voltage source applies voltage to the defect model through a 5kΩ protection resistor, and the induced partial discharge signal X[n] and the voltage source signal in the UHF antenna are collected by an oscilloscope and a high-voltage differential probe, respectively.

[0034] In the embodiment of the present invention, in step 2, the collected signal is decomposed by Daubechies-2 (db2) wavelet, the energy of each layer of detail components is calculated by the following formula, the number of layers L of the detail components with the minimum energy is determined, and the white noise is reconstructed with all detail components with a number of layers less than or equal to L.

[0035]

[0036] Among them, X j [i] is the j-th layer detail component of the original discharge signal; Function Returns The j corresponding to the maximum value.

[0037] In the embodiment of the present invention, the energy ratio of the pure signal and the reconstructed white noise in each high-frequency signal layer is calculated in step 3, and the layer with the largest energy ratio is used as the decomposition layer number of the collected signal:

[0038]

[0039] Among them, R is the number of wavelet decomposition layers; X j [i] is the j-th layer detail component of the original discharge signal; N j [l] is the j-th layer detail component of the noise signal.

[0040] In the embodiment of the present invention, in step 4, the partial discharge signal X[n] is subjected to R-layer wavelet decomposition, and a multi-scale threshold function is designed to perform threshold processing on the wavelet coefficients of each layer. The multi-scale threshold function is as follows:

[0041]

[0042] Among them, T0 is the adaptive threshold, and the calculation formula is: X j [i] is the detail component of the jth layer of the partial discharge signal; |X j [i]| med is the median value of the detail component; L(X j [i]) is the length of the detail component.

[0043] In the embodiment of the present invention, in step 5, N sampling point data between voltage zero crossing and stable state are taken as potential aliasing signal s, and the nth signal s n and the n+1th signal s n+1 The relationship coefficient between them is:

[0044]

[0045] Among them, s n [i]、s n+1 [i] is s n 、s n+1 The i-th sampling point of For n 、s n+1 The average value of .

[0046] In the embodiment of the present invention, s1 is used as the reference signal to calculate the correlation coefficients of all potential aliasing signals. If the correlation coefficients are all greater than 0.95, it can be considered that all potential aliasing signals are power supply interference and can be reset to zero; if the correlation coefficient is less than or equal to 0.95 and the oscillation is severe, it is considered that the power supply interference signal is aliased with the local discharge signal, and the collected signal is filtered by a high-pass filter. Reduce the pulse voltage to below the discharge voltage, take the cutoff frequency of the test signal with a voltage rise time of 1μs as the reference, and design the cutoff frequency of the high-pass filter under different voltage rise times by the following formula:

[0047]

[0048] Among them, f cb is the cut-off frequency of the test signal when the voltage rise time is 1μs; T a is the voltage rise time.

[0049] The implementation cases of the present invention are as follows:

[0050] First, build Figure 2The test platform shown in the figure, the ZMC-100 high-frequency pulse voltage source in the platform applies voltage to the cable defect through a 5kΩ protection resistor, and the partial discharge signal and voltage source signal induced in the UHF antenna are collected by a digital oscilloscope and a high-voltage differential probe respectively. The test voltage amplitude is 1.5 times the initial discharge voltage, and the frequency is a bipolar square wave voltage of 1kHz. After the discharge is stable, the discharge signal is collected, and a section of the discharge signal is taken as Figure 3 As shown; secondly, the discharge signal is decomposed by wavelet, and three layers of wavelet detail components are used to reconstruct white noise according to the energy distribution; then, by calculating the energy ratio of noise to discharge signal under different decomposition layers, the optimal decomposition layer is obtained to be 5 layers, and the sample signal is denoised by multi-scale wavelet threshold denoising method; finally, the signal between voltage zero crossing and stability is taken as the potential aliasing signal, and the mutual correlation coefficient is used to determine whether the signal has edge overshoot aliasing. It is found that the coefficients are all greater than 0.95, so there is no edge overshoot aliasing, and the potential aliasing signal is directly removed to obtain the final effect as shown Figure 4 shown.

Claims

1. A method for extracting insulation partial discharge signals under high-frequency pulse voltage, characterized in that: The steps of the identification method include: Step 1: Build a partial discharge test platform under high-frequency pulses and collect partial discharge signals through an ultra-high frequency antenna; Step 2: Perform wavelet decomposition on the collected partial discharge signal and reconstruct the white noise according to the energy distribution of the decomposed signal; Step 3: Determine the decomposition level of the collected signal by the energy ratio of the pure signal and the reconstructed white noise at each wavelet scale; Step 4: According to the number of decomposition layers, design a multi-scale threshold function to remove signal noise in layers; Step 5: After removing the signal noise, take the signal between voltage zero crossing and stability as the potential aliasing signal, and use the cross-correlation coefficient to determine whether the signal has edge overshoot aliasing. If the information is not aliased, directly remove the power supply interference. Otherwise, design a high-pass filter based on the pulse edge to eliminate the edge overshoot interference.

2. A method for extracting insulation partial discharge signals under high-frequency pulse voltage according to claim 1, characterized in that: In step 2, the collected partial discharge signal is decomposed by Dobesie No. 2 db2 wavelet, and the energy of each layer of detail components is calculated by the following formula, the number of layers L of the detail component with the minimum energy is determined, and all detail components with the number of layers less than or equal to L are used to reconstruct white noise: Among them, X j [i] is the j-th layer detail component of the original discharge signal; Function Returns The j corresponding to the maximum value.

3. A method for extracting insulation partial discharge signals under high-frequency pulse voltage according to claim 1, characterized in that: In step 3, the number of decomposition layers is determined according to the energy ratio of the pure signal and the reconstructed white noise in each high-frequency signal layer: Among them, R is the number of wavelet decomposition layers; X j [i] is the j-th layer detail component of the original discharge signal; N j [l] is the j-th layer detail component of the reconstructed white noise.

4. A method for extracting insulation partial discharge signals under high-frequency pulse voltage according to claim 1, characterized in that: Step 4 includes, The local discharge signal X[n] is decomposed by R-layer wavelet, and a multi-scale threshold function is designed to perform threshold processing on the wavelet coefficients of each layer. The multi-scale threshold function is as follows: Among them, T0 is the adaptive threshold, and the calculation formula is: X j [i] is the detail component of the jth layer of the partial discharge signal; |X j [i]| med is the median value of the detail component; L(X j [i]) is the length of the detail component.

5. A method for identifying insulation partial discharge patterns under small sample conditions according to claim 1, characterized in that: Step 5 includes, Design a program to determine whether the power supply interference information is aliased with the partial discharge information; take the N sampling point data between the voltage zero crossing and the stable state as the potential aliasing signal s n , taking the first s1 as the reference signal, calculate the correlation coefficients of all potential aliasing signals; if the correlation coefficients are all greater than 0.95, it can be considered that the potential aliasing signals are all power supply interference at this time, and can be set to zero; if the correlation coefficient is less than or equal to 0.95 and the oscillation is severe, it is considered that the power supply interference signal is aliased with the partial discharge signal at this time, and the collected signal is filtered through a high-pass filter; reduce the pulse voltage to below the discharge voltage, and take the cutoff frequency of the test signal with a voltage rise time of 1μs as the reference, and design the cutoff frequency of the high-pass filter under different voltage rise times by the following formula: Among them, f cb is the cut-off frequency of the test signal when the voltage rise time is 1μs; T a is the voltage rise time.

Images