Discharge identification method under overhead lines

Abstract

The present invention discloses a method for identifying discharges under overhead lines, which relates to the technical field of electrical signal processing. The method comprises: obtaining a zero-sequence voltage signal of an overhead line before a wildfire discharge, and performing low-pass filtering preprocessing on the zero-sequence voltage signal; performing continuous wavelet transform on the preprocessed zero-sequence voltage signal to obtain time-domain wavelet coefficients; performing phase analysis on the wavelet coefficients at each time position, and determining the instantaneous frequency by using the phase change of the wavelet coefficients; synchronously squeezing the wavelet coefficients according to the instantaneous frequency, and redistributing the wavelet coefficients to corresponding frequency positions to obtain a time-frequency diagram of the zero-sequence voltage; extracting the main frequency component, frequency bandwidth, energy distribution and spectrum peak position of the low-frequency band from the time-frequency diagram to construct characteristic parameters, judging the type of the zero-sequence voltage signal, and thereby determining the cause of the wildfire accident; the present invention uses the zero-sequence voltage signal as a detection signal, which is more conducive to investigators judging the cause of the wildfire accident.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical signal processing, in particular to a discharge identification method under an overhead line. Background Art

[0002] In recent years, 10kV line tripping accidents have occurred frequently due to social factors and tree faults on 10kV overhead lines that caused wildfires. Scientifically, rapidly, and accurately warning of wildfires and tree faults, as well as analyzing and determining the causes of these accidents, are of great technical value and socioeconomic benefit.

[0003] The synchronized squeezed wavelet transform (SSWT) is a common technique for signal time-frequency analysis, widely used for feature extraction and pattern recognition of complex signals. Flame discharge and treeline discharge signals typically have different frequency characteristics and are often affected by noise. The synchronized squeezed wavelet transform calculates the instantaneous frequency of the signal through the continuous wavelet transform (CWT) and redistributes the wavelet coefficients in the frequency domain to generate a high-precision time-frequency plot, from which rich spectral information can be extracted. However, the recognition effectiveness of the SSWT depends heavily on the selection of the mother wavelet and the accurate calculation of the instantaneous frequency. The Morlet wavelet, due to its excellent time and frequency localization properties, is often used in the analysis of flame and treeline discharge signals.

[0004] When identifying wildfire flame discharges and treeline discharges, the zero-sequence voltage signal contains a large number of time-varying frequency components. Flame discharge signals tend to have a wide spectral distribution and large energy fluctuations, while the spectral distribution of treeline discharge signals is relatively concentrated. Using the synchronized squeezing wavelet transform, the energy of the zero-sequence voltage signal can be redistributed on the time-frequency plane, concentrating the frequency components at the corresponding instantaneous frequencies. This improves frequency resolution and enables a clear representation of signal characteristics. Summary of the Invention

[0005] In order to solve the problems existing in the prior art, the purpose of the present invention is to provide a discharge identification method under overhead lines. The present invention is used to investigate the cause of wildfire accidents under 10kV overhead lines. The present invention uses zero-sequence voltage signals as detection signals, which is more conducive to investigators to determine the cause of wildfire accidents.

[0006] To achieve the above objectives, the present invention adopts a technical solution: a method for identifying discharges under overhead lines, which is used to investigate the causes of wildfire accidents under overhead lines. The causes of the accidents include wildfire discharges caused by social factors and wildfire discharges caused by tree line faults. The method comprises the following steps:

[0007] Step 1: obtaining a zero-sequence voltage signal of an overhead line before a wildfire discharge, and performing low-pass filtering preprocessing on the zero-sequence voltage signal;

[0008] Step 2: Select Morlet wavelet as the mother wavelet function ψ(t), perform continuous wavelet transform on the pre-processed zero-sequence voltage signal, and obtain the time domain wavelet coefficient W x (a,b);

[0009] Step 3: Perform phase analysis on the wavelet coefficients at each time position b, and use the phase change of the wavelet coefficients to determine the instantaneous frequency ω x (a,b);

[0010] Step 4: Synchronously squeeze the wavelet coefficients according to the instantaneous frequency, redistribute the wavelet coefficients to the corresponding frequency positions, and obtain the time-frequency diagram of the zero-sequence voltage;

[0011] Step 5: Extract the main frequency component, frequency bandwidth, energy distribution, and spectrum peak position of the low-frequency band from the time-frequency diagram to construct characteristic parameters, determine the type of zero-sequence voltage signal, and thus determine the cause of the wildfire accident.

[0012] As a further improvement of the present invention, in step 2, the mother wavelet function ψ(t) is specifically as follows:

[0013]

[0014] Where, is a normalization factor, exp(jω0t) is a complex exponential term, representing a sine wave with a frequency of ω0. It is a Gaussian envelope term, which is used to confine the sine wave to a limited time range and produce a localized effect;

[0015] The continuous wavelet transform is as follows:

[0016]

[0017] Where a is the scale parameter, b is the time position, and is the translation parameter. Used for signal energy normalization, x(t) represents the zero-sequence voltage signal after preprocessing, Represents the conjugate function of the mother wavelet function, W x (a, b) represents the energy intensity of the preprocessed zero-sequence voltage signal x(t) at scale parameter a and time position b, that is, the wavelet coefficient.

[0018] As a further improvement of the present invention, in step 3, the instantaneous frequency is specifically as follows:

[0019]

[0020] Where, Represents the partial derivative of the wavelet coefficient with respect to the time position b, i represents the imaginary part, Wx (a,b) are wavelet coefficients, ω x (a,b) represents the instantaneous frequency of the signal at scale parameter a and time position b.

[0021] As a further improvement of the present invention, in step 4, the wavelet coefficients are synchronously squeezed according to the instantaneous frequency as follows:

[0022]

[0023] Where, T x (f,b) represents the time-frequency distribution of the synchronous squeezing wavelet transform, δ(f-ω x (a,b)) is the Dirac function used to redistribute the wavelet coefficients W in the frequency domain x The energy of (a,b), Represents the normalization factor of the scale parameter a.

[0024] As a further improvement of the present invention, the step 5 is specifically as follows:

[0025] Based on the main frequency components, frequency bandwidth, energy distribution and spectrum peak position, the corresponding feature vectors of wildfire discharges caused by social reasons and wildfire discharges caused by tree-line faults are constructed; the feature vectors corresponding to each feature parameter are input into the trained support vector machine. If the output result meets the discharge characteristics of wildfires caused by social reasons, it is determined that the wildfire was caused by social reasons and the cause of the accident is not related to the overhead lines; if the output result meets the discharge characteristics of wildfires caused by tree-line faults, it is determined that the wildfire was caused by tree-line faults and the cause of the accident is related to the overhead lines.

[0026] As a further improvement of the present invention, in step 5, the main frequency component f min The details are as follows:

[0027]

[0028] Where, T x (f,b) represents the time-frequency of the synchronized squeezed wavelet transform, It means finding the frequency with the highest energy density.

[0029] As a further improvement of the present invention, in step 5, the energy distribution E(f) is obtained by calculating the energy density in different frequency bands:

[0030]

[0031] Where b1 and b2 are the starting and ending points of the interval in time, and E(f) represents the total energy of the signal at frequency f. By normalizing the energy density of different frequency bands, the relative dispersion of the energy distribution is obtained:

[0032]

[0033] Where, Indicates the average frequency.

[0034] As a further improvement of the present invention, the overhead line is a 10kV overhead line.

[0035] The beneficial effects of the present invention are:

[0036] The present invention distinguishes between the two situations where the frequency of the zero-sequence voltage signal of the 10kV overhead line before the wildfire is caused by social reasons and that caused by tree-line faults is in the low-frequency region below 100Hz. By extracting the characteristic vectors corresponding to the characteristic parameters of the main frequency component, frequency bandwidth, energy distribution and spectrum peak of the zero-sequence voltage signal in the low-frequency band, the signal type is judged based on the trained vector machine, thereby determining the type of wildfire accident, facilitating the rapid and accurate judgment of the cause of the accident after the wildfire occurs. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Flowchart of an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0039] Example

[0040] like Figure 1 As shown, a method for identifying discharges under overhead lines is used to investigate the causes of wildfire accidents under overhead lines. The causes of the accidents include wildfire discharges caused by social factors and wildfire discharges caused by tree line faults. The method comprises the following steps:

[0041] Step 1: Obtain the 10kV overhead line zero-sequence voltage signal before the wildfire caused by social reasons and tree-line faults, and perform low-pass filtering preprocessing on the zero-sequence voltage signal, as follows:

[0042] The lowpass library function in MATLAB is used for low-pass filtering. After studying the zero-sequence voltage frequency in the low-frequency region below 100Hz when wildfire flame discharge and treeline discharge occur under 10kV overhead lines, it is found that there is a significant difference between the two situations. Therefore, the zero-sequence voltage signal outside 100Hz is filtered out to obtain the pre-processed zero-sequence voltage signal.

[0043] Step 2: Select Morlet wavelet as the mother wavelet function ψ(t), perform continuous wavelet transform on the pre-processed zero-sequence voltage signal, and obtain the time domain wavelet coefficient W x (a,b);

[0044] Step 3: Perform phase analysis on the wavelet coefficients at each time position b, and use the phase change of the wavelet coefficients to determine the instantaneous frequency ω x (a,b);

[0045] Step 4: Synchronously squeeze the wavelet coefficients according to the instantaneous frequency, redistribute the wavelet coefficients to the corresponding frequency positions, and obtain the time-frequency of the zero-sequence voltage;

[0046] Step 5: Extract the main frequency component, frequency bandwidth, energy distribution, and spectrum peak position of the low-frequency band from the time-frequency diagram to construct characteristic parameters, determine the type of zero-sequence voltage signal, and thus determine the cause of the wildfire accident;

[0047] In step 2, the Morlet wavelet is selected as the mother wavelet function ψ(t), and the pre-processed zero-sequence voltage signal is subjected to continuous wavelet transform to obtain the time domain wavelet coefficient W x (a, b), the formula for continuous wavelet transform is:

[0048]

[0049] Where a is the scale parameter, take a>1, b is the translation parameter, Used for signal energy normalization, x(t) represents the zero-sequence voltage signal after preprocessing, Represents the conjugate function of the mother wavelet function, W x (a, b) represents the energy intensity of the zero-sequence voltage signal x(t) at scale parameter a and time position b.

[0050] In step 3, the phase of the wavelet coefficients at each time position b is analyzed, and the instantaneous frequency is determined by the phase change of the wavelet coefficients. The instantaneous frequency ω x The formula for (a,b) is:

[0051]

[0052] Where, Represents the partial derivative of the wavelet coefficient with respect to time b, i represents the imaginary part, W x (a,b) are the wavelet coefficients obtained in step 2, ω x (a,b) represents the instantaneous frequency of the signal at scale parameter a and time position b.

[0053] In step 4, the wavelet coefficients are synchronously squeezed according to the instantaneous frequency, and the wavelet coefficients are redistributed to the corresponding frequency positions to obtain the time-frequency diagram of the zero-sequence voltage. The synchronous squeezing wavelet transform calculation formula is:

[0054]

[0055] Where, T x (f,b) represents the time-frequency distribution of the synchronous squeezing wavelet transform, δ(f-ω x (a,b)) is the Dirac function used to redistribute the wavelet coefficients W in the frequency domain x The energy of (a,b), Represents the normalization factor of scale a.

[0056] In step 5, the characteristic vectors corresponding to flame discharge and treeline discharge are constructed based on the main frequency components, frequency bandwidth, energy distribution, and spectrum peak position. The characteristic vectors corresponding to each characteristic parameter are input into the trained support vector machine. If the output result meets the flame discharge characteristics, it is determined that the wildfire was caused by social reasons and the cause of the accident is not related to the 10kV overhead line. If it meets the treeline discharge characteristics, it is determined that the wildfire was caused by a treeline fault and the cause of the accident is related to the 10kV overhead line.

[0057] In step 5, the main frequency component, frequency bandwidth, energy distribution and spectrum peak position of the low frequency band are extracted from the time-frequency graph to construct characteristic parameters. On the time-frequency graph, for each time position b, the main frequency component f min The calculation formula is as follows:

[0058]

[0059] Where, T x (f,b) represents the time-frequency of the synchronized squeezed wavelet transform, It means finding the frequency with the highest energy density.

[0060] In step 5, the main frequency component, frequency bandwidth, energy distribution, and spectral peak position of the low-frequency band are extracted from the time-frequency graph to construct characteristic parameters. At a given time position b, the energy distribution E(f) can be obtained by calculating the energy density in different frequency bands. The calculation formula for energy distribution E(f) is as follows:

[0061]

[0062] Where b1 and b2 are the starting and ending points of the time interval, and E(f) represents the total energy of the signal at frequency f. By normalizing the energy density of different frequency bands, the relative dispersion of the energy distribution can be obtained. The calculation formula for the relative dispersion is as follows:

[0063]

[0064] Where E(f) represents the total energy of the signal at frequency f, mean represents the average frequency and is defined as follows:

[0065]

[0066] Where E(f) represents the total energy of the signal at frequency f.

[0067] In this embodiment, the zero-sequence voltage signal of the 10kV overhead line before the wildfire caused by social reasons and tree-line fault is first obtained, and the zero-sequence voltage signal is pre-processed by low-pass filtering; the Morlet wavelet is selected as the mother wavelet function ψ(t), and the pre-processed zero-sequence voltage signal is subjected to continuous wavelet transform to obtain the time domain wavelet coefficient W x (a, b); perform phase analysis on the wavelet coefficients at each time position b, and use the phase change of the wavelet coefficients to determine the instantaneous frequency; synchronously squeeze the wavelet coefficients according to the instantaneous frequency and redistribute them to the corresponding frequency positions to obtain a time-frequency diagram of the zero-sequence voltage; extract the main frequency component, frequency bandwidth, energy distribution, and spectral peak position of the low-frequency band from the time-frequency diagram to construct characteristic parameters. The eigenvectors corresponding to each characteristic parameter are input into a trained support vector machine to determine the type of zero-sequence voltage signal and thus determine the cause of the wildfire accident. This improves the efficiency and accuracy of determining the cause of the wildfire accident.

[0068] This embodiment has the following advantages: in actual situations, the evidence used to investigate the cause of a wildfire is basically destroyed. The zero-sequence voltage signal, as one of the signals collected by power companies on a long-term and fixed basis, has the advantages of being stable and easy to obtain. Feature parameters are constructed based on the main frequency component, frequency bandwidth, energy distribution, and spectrum peak position of the zero-sequence voltage in the low-frequency band. The feature vectors corresponding to the feature parameters are input into a trained support vector machine to determine the type of the zero-sequence voltage signal. This can effectively avoid errors in human subjective judgment and is very convenient for determining the cause of an accident after a wildfire.

[0069] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A method for identifying discharges under overhead lines, characterized in that: The method is used to investigate the causes of wildfire accidents under overhead lines, wherein the causes of the accidents include wildfire discharges caused by social factors and wildfire discharges caused by tree line faults, and the method comprises the following steps: Step 1: obtaining a zero-sequence voltage signal of an overhead line before a wildfire discharge, and performing low-pass filtering preprocessing on the zero-sequence voltage signal; Step 2: Select Morlet wavelet as the mother wavelet function ψ(t), perform continuous wavelet transform on the pre-processed zero-sequence voltage signal, and obtain the time domain wavelet coefficient W x (a,b); In step 2, the mother wavelet function ψ(t) is specifically as follows: Where, is a normalization factor, exp(jω0t) is a complex exponential term, representing a sine wave with a frequency of ω0. It is a Gaussian envelope term, which is used to confine the sine wave to a limited time range and produce a localized effect; The continuous wavelet transform is as follows: Where a is the scale parameter, b is the time position, and is the translation parameter. Used for signal energy normalization, x(t) represents the zero-sequence voltage signal after preprocessing, Represents the conjugate function of the mother wavelet function, W x (a, b) represents the energy intensity of the preprocessed zero-sequence voltage signal x(t) at scale parameter a and time position b, i.e., the wavelet coefficient; Step 3: Perform phase analysis on the wavelet coefficients at each time position b, and use the phase change of the wavelet coefficients to determine the instantaneous frequency ω x (a,b); In step 3, the instantaneous frequency is as follows: Where, Represents the partial derivative of the wavelet coefficient with respect to the time position b, i represents the imaginary part, W x (a,b) are wavelet coefficients, ω x (a,b) represents the instantaneous frequency of the signal at scale parameter a and time position b; Step 4: Synchronously squeeze the wavelet coefficients according to the instantaneous frequency, redistribute the wavelet coefficients to the corresponding frequency positions, and obtain the time-frequency diagram of the zero-sequence voltage; Step 5: Extract the main frequency component, frequency bandwidth, energy distribution, and spectrum peak position of the low-frequency band from the time-frequency diagram to construct characteristic parameters, determine the type of zero-sequence voltage signal, and thus determine the cause of the wildfire accident; In step 5, the main frequency component f min The details are as follows: Where, T x (f,b) represents the time-frequency of the synchronized squeezed wavelet transform, It means finding the frequency with the largest energy density; In step 5, the energy distribution E(f) is obtained by calculating the energy density in different frequency bands: Where b1 and b2 are the starting and ending points of the interval in time, and E(f) represents the total energy of the signal at frequency f. By normalizing the energy density of different frequency bands, the dispersion of the energy distribution is obtained: Where, Indicates the average frequency.

2. The method for identifying discharge under overhead lines according to claim 1, characterized in that: In step 4, the wavelet coefficients are synchronously squeezed according to the instantaneous frequency as follows: Where, T x (f,b) represents the time-frequency distribution of the synchronous squeezing wavelet transform, δ(f-ω x (a,b)) is the Dirac function used to redistribute the wavelet coefficients W in the frequency domain x The energy of (a,b), Represents the normalization factor of the scale parameter a.

3. The method for identifying discharge under overhead lines according to claim 1, characterized in that: The step 5 is specifically as follows: Based on the main frequency components, frequency bandwidth, energy distribution, and spectrum peak positions, feature vectors corresponding to wildfire discharges caused by social factors and wildfire discharges caused by treeline faults are constructed. The feature vectors corresponding to each feature parameter are input into a trained support vector machine. If the output meets the characteristics of wildfire discharge caused by social factors, it is determined that the wildfire was caused by social factors and the cause of the accident was not related to overhead lines. If the output results meet the discharge characteristics of wildfires caused by tree-line faults, it is determined that the wildfire was caused by tree-line faults and the cause of the accident is related to overhead lines.

4. The method for identifying discharge under an overhead line according to any one of claims 1 to 3, characterized in that: The overhead line is a 10kV overhead line.

Images