CN-121995162-A - Transformer partial discharge positioning detection method based on ultrahigh frequency signals

Abstract

A transformer partial discharge positioning detection method based on an ultrahigh frequency signal is characterized by comprising the following steps of S1, estimating time delay of signals obtained by an ultrahigh frequency (UHF) sensor array based on a generalized cross-correlation method. S2, the signals of the UHF sensor array are arranged and combined, and positioning point coordinates are solved based on an improved multi-moth optimization (MFO) algorithm according to the time delay obtained in the step S1. S3, clustering the plurality of positioning point coordinates obtained in the step S2 based on a density clustering algorithm (DBSCAN), and selecting the geometric center position of the category with the largest sample number as the final partial discharge source coordinate.

Inventors

• CAI QIAN
• SHENG GEHAO
• QIAN YONG
• XU ZHIREN
• CHEN ZEHAO
• LI JIAYANG
• WANG HUI
• SHU BO
• LUO LINGEN
• SONG HUI

Assignees

• 上海交通大学

Dates

Publication Date

20260508

Application Date

20241105

Claims (6)

1. 1. The transformer partial discharge positioning detection method based on the ultrahigh frequency signal is characterized by comprising the following steps of: s1, estimating the time delay of signals obtained by an Ultra High Frequency (UHF) sensor array based on a generalized cross-correlation method, wherein the method specifically comprises the following steps: s1.1, obtaining an autocorrelation power spectrum of a first signal x1 (t); S1.2, a cross-correlation function of a first signal x1 (t) and a second signal x2 (t) is obtained; S1.3, performing secondary correlation and weighting treatment on the autocorrelation power spectrum and the cross correlation function to obtain a weighted cross power spectrum function; s1.4, carrying out inverse Fourier transform on the weighted cross power spectrum function to obtain a time delay function, and taking the peak value of the function as the required time delay; S2, arranging and combining signals of the UHF sensor array, and solving positioning point coordinates based on an improved multi-moth optimization (MFO) algorithm according to the time delay obtained in the step S1, wherein the method specifically comprises the following steps: S2.1, signals of M-1 UHF sensor arrays are arranged and combined to obtain L combination results; S2.2, calculating positioning point coordinates of each combined result by adopting an improved MFO algorithm; S3, clustering the plurality of positioning point coordinates obtained in the step S2 based on a density clustering algorithm (DBSCAN), and selecting the geometric center position of the category with the largest sample number as the final partial discharge source coordinate.
2. 2. The method for detecting partial discharge positioning of a transformer based on an ultrahigh frequency signal according to claim 1, wherein in the step S1.1, a frequency domain signal obtained after fourier transform of a first signal x1 (t) and a conjugate thereof are used when an autocorrelation power spectrum is obtained, and in the step S1.2, a conjugate of a frequency domain signal obtained after fourier transform of a second signal x2 (t) is used when a cross correlation function is obtained.
3. 3. The method for detecting partial discharge positioning of a transformer based on an uhf signal according to claim 1, wherein the modified MFO algorithm in step S2.2 specifically comprises: S2.2.1 randomly generating an initial moth population, wherein the number of the moth populations and the dimension of a solving problem variable are determined according to the distance range of the inner wall of the transformer; S2.2.2 calculating the fitness of each moth population to obtain a fitness matrix of the moth population; S2.2.3 sorting the moth populations according to the fitness to form flame populations, and calculating a fitness matrix of the flame populations; S2.2.4 in the iterative process, the moths move to the flame along the logarithmic spiral curve to find a better solution; s2.2.5 in the iterative process, flame population is adaptively reduced.
4. 4. The method for detecting partial discharge positioning of a transformer based on an ultrahigh frequency signal according to claim 1, wherein the DBSCAN algorithm in step S3 specifically comprises: s3.1, initializing parameters including an input sample set, a neighborhood radius epsilon and a neighborhood density threshold MinPts; s3.2, finding out a core object, namely, satisfying sample points with the number of samples in the neighborhood being more than or equal to MinPts; S3.3, expanding the cluster by taking the core object as a starting point until a cluster ending condition is met; s3.4, outputting a clustering result, and selecting the geometric center position of the cluster with the largest sample number as the final partial discharge source coordinate.
5. 5. The method for detecting partial discharge positioning of a transformer based on an ultrahigh frequency signal according to any one of claims 1 to 4, further comprising optimizing a layout of the sensor array to improve positioning accuracy.
6. 6. The method for localized discharge detection of uhf signal-based transformers of any of claims 1-4, further comprising pre-processing the uhf signal to reduce noise interference.

Description

Transformer partial discharge positioning detection method based on ultrahigh frequency signals Technical Field The invention belongs to the field of equipment insulation defect detection, and relates to a transformer partial discharge positioning detection method based on an ultrahigh frequency signal. Background The degree of insulation degradation of a transformer has been an important indicator for measuring the health level of the transformer. When the transformer is affected by a defective production of equipment or a severe environment during transportation, air gap bubbles are generated in the insulation structure, resulting in insulation degradation, partial discharge phenomenon is easily generated inside the transformer. Numerous previous studies have shown that the degree of insulation degradation of a transformer has a direct relationship with the location of the partial discharge source. In addition, once the field maintenance personnel detect that partial discharge exists in the transformer, the accurate positioning of the position of the discharge source is important, because the position is directly related to the establishment and implementation efficiency of the follow-up maintenance strategy. Therefore, the research of the transformer partial discharge positioning technology has important significance for transformer equipment state monitoring and fault diagnosis. When partial discharge occurs, various physical signals such as light, ultrasonic waves, ultra High Frequency (UHF) electromagnetic waves and the like are generated, and corresponding partial discharge detection methods such as a photodetection method, an ultrasonic method and an ultra high frequency method are adopted, wherein the ultra high frequency method is outstanding in the field of partial discharge detection of transformers due to the excellent anti-interference capability and high sensitivity, and becomes a hot spot of current research. The method realizes high-precision positioning of the partial discharge source by capturing the ultrahigh frequency electromagnetic wave signal in the range of 300 to 3000MHz generated in the transformer. Therefore, the partial discharge positioning technology of the transformer based on the ultrahigh frequency signal is a key content of partial discharge detection. Disclosure of Invention The invention aims to improve the precision of the partial discharge positioning of a transformer, and provides a method for detecting the partial discharge positioning of the transformer based on a ultrahigh frequency signal, which realizes the time delay estimation, the positioning equation solving and the positioning point correcting of the partial discharge ultrahigh frequency signal of the transformer by combining three algorithms, namely a generalized secondary cross-correlation (GSCC), an improved moth fire suppression algorithm (MFO) and a density clustering algorithm (DBSCAN). In order to achieve the above purpose, the present invention adopts the following technical scheme: The transformer partial discharge positioning detection method based on the ultrahigh frequency signal is characterized by comprising the following steps of: s1, estimating the time delay of signals obtained by an Ultra High Frequency (UHF) sensor array based on a generalized cross-correlation method, wherein the method specifically comprises the following steps: s1.1, obtaining an autocorrelation power spectrum of a first signal x1 (t); S1.2, a cross-correlation function of a first signal x1 (t) and a second signal x2 (t) is obtained; S1.3, performing secondary correlation and weighting treatment on the autocorrelation power spectrum and the cross correlation function to obtain a weighted cross power spectrum function; s1.4, carrying out inverse Fourier transform on the weighted cross power spectrum function to obtain a time delay function, and taking the peak value of the function as the required time delay; S2, arranging and combining signals of the UHF sensor array, and solving positioning point coordinates based on an improved multi-moth optimization (MFO) algorithm according to the time delay obtained in the step S1, wherein the method specifically comprises the following steps: S2.1, signals of M-1 UHF sensor arrays are arranged and combined to obtain L combination results; S2.2, calculating positioning point coordinates of each combined result by adopting an improved MFO algorithm; S3, clustering the plurality of positioning point coordinates obtained in the step S2 based on a density clustering algorithm (DBSCAN), and selecting the geometric center position of the category with the largest sample number as the final partial discharge source coordinate. Further, in the step S1.1, the frequency domain signal obtained after the Fourier transform of the first signal x1 (t) and the conjugate thereof are used when the autocorrelation power spectrum is obtained, and in the step S1.2, the conjugate of the frequency domai