CN-122017485-A - Partial discharge real-time monitoring method based on ultrahigh frequency signal and ultrasonic signal

CN122017485ACN 122017485 ACN122017485 ACN 122017485ACN-122017485-A

Abstract

The invention relates to the technical field of discharge monitoring, and discloses a partial discharge real-time monitoring method based on an ultrahigh frequency signal and an ultrasonic signal, which comprises the steps of acquiring an original monitoring data set synchronously acquired when tested power equipment operates, wherein the original monitoring data set comprises an ultrahigh frequency original signal sequence and an ultrasonic original signal sequence; the method comprises the steps of processing an original monitoring data set to obtain a standard monitoring data set, extracting waveform characteristics of the ultra-high frequency purifying signal sequence to construct an ultra-high frequency characteristic vector set, extracting sound field characteristics of the ultrasonic purifying signal sequence to construct an ultrasonic characteristic vector set, generating a combined characteristic matrix based on the ultra-high frequency characteristic vector set and the ultrasonic characteristic vector set, constructing a partial discharge identification model according to the combined characteristic matrix to obtain a partial discharge real-time diagnosis monitoring result of the tested power equipment, guaranteeing the real-time diagnosis monitoring precision of the partial discharge, and improving the early identification accuracy and the predictive maintenance level of equipment insulation defects.

Inventors

LU SINAN
GAO CHAO
ZHANG XIAOLONG
SHI YINGCAI
RUAN YIXIAO
Liao Ruocen
ZHOU YANG
ZHANG YILIN

Assignees

国网重庆市电力公司超高压分公司

Dates

Publication Date: 20260512
Application Date: 20260126

Claims (10)

1. The partial discharge real-time monitoring method based on the ultrahigh frequency signal and the ultrasonic signal is characterized by comprising the following steps of: acquiring an original monitoring data set synchronously acquired when the tested power equipment operates, wherein the original monitoring data set comprises an ultrahigh frequency original signal sequence and an ultrasonic original signal sequence; performing signal purification treatment on the original monitoring data set to obtain a standard monitoring data set, wherein the standard monitoring data set comprises an ultrahigh frequency purification signal sequence and an ultrasonic purification signal sequence; extracting waveform characteristics of the ultrahigh frequency purification signal sequence, constructing an ultrahigh frequency characteristic vector set, extracting sound field characteristics of the ultrasonic purification signal sequence, and constructing an ultrasonic characteristic vector set; based on the ultrahigh frequency feature vector set and the ultrasonic feature vector set, carrying out multi-mode information fusion to generate a joint feature matrix; And constructing a partial discharge identification model according to the joint feature matrix to obtain a real-time diagnosis and monitoring result of the partial discharge of the tested power equipment.
2. The method for monitoring partial discharge in real time based on ultrahigh frequency signals and ultrasonic signals according to claim 1, wherein when acquiring an original monitoring data set synchronously acquired during operation of the tested power equipment, the method comprises the following steps: controlling an ultrahigh frequency sensor probe and an ultrasonic sensor probe to acquire signals under the same time reference through a synchronous trigger mechanism to obtain an initial ultrahigh frequency signal segment and an initial ultrasonic signal segment; performing time stamp alignment processing on the initial ultrahigh frequency signal segment and the initial ultrasonic signal segment to generate a time synchronization signal pair; and integrating all time synchronization signal pairs in the acquisition period into the original monitoring data set.
3. The method for monitoring partial discharge in real time based on ultrahigh frequency signals and ultrasonic signals according to claim 1, wherein when the original monitoring data set is subjected to signal purification treatment, the method for monitoring partial discharge in real time based on ultrahigh frequency signals and ultrasonic signals comprises the following steps: carrying out noise base treatment on each ultrahigh frequency original signal in the ultrahigh frequency original signal sequence to obtain a noise power level; constructing an adaptive filter according to the noise power level, and performing narrow-band interference suppression on the ultrahigh frequency original signal to obtain an ultrahigh frequency filtering signal; Carrying out waveform amplitude normalization processing on the ultrahigh frequency filtered signal to map a signal peak to a unified dimension interval so as to obtain the ultrahigh frequency purified signal sequence; collecting environmental noise of each ultrasonic original signal in the ultrasonic original signal sequence, and obtaining a background sound field sample; Constructing a sound field cancellation model based on the background sound field sample, and performing environmental noise cancellation processing on the ultrasonic original signal to obtain an ultrasonic noise reduction signal; and performing sound intensity calibration processing on the ultrasonic noise reduction signal to normalize the sound pressure amplitude value to the same reference dimension, thereby obtaining the ultrasonic purification signal sequence.
4. The method for monitoring partial discharge in real time based on an ultrahigh frequency signal and an ultrasonic signal according to claim 1, wherein when extracting waveform characteristics of the ultrahigh frequency purifying signal sequence and constructing an ultrahigh frequency characteristic vector set, the method comprises the following steps: performing time-frequency conversion treatment on each ultrahigh frequency purifying signal in the ultrahigh frequency purifying signal sequence to generate an ultrahigh frequency time-frequency spectrogram; Extracting spectrogram statistical characteristics from the ultrahigh frequency time-frequency spectrogram, wherein the spectrogram statistical characteristics comprise spectral energy distribution concentration degree, a main frequency offset index and a high frequency component attenuation coefficient; Performing pulse envelope detection on the ultrahigh frequency purification signal, and extracting pulse waveform parameters including rising edge steepness, pulse width variation coefficient and half-wave duration; integrating the spectrogram statistical features and the pulse waveform parameters into a partial discharge fingerprint feature tuple; And aggregating fingerprint characteristic tuples corresponding to all the ultrahigh frequency purification signals into the ultrahigh frequency characteristic vector set.
5. The method for monitoring partial discharge in real time based on ultrahigh frequency signals and ultrasonic signals according to claim 1, wherein when extracting sound field characteristics of the ultrasonic purification signal sequence and constructing an ultrasonic characteristic vector set, comprising: Performing wavelet decomposition on each ultrasonic purification signal in the ultrasonic purification signal sequence to obtain a multi-scale sound field component; Extracting energy distribution characteristics from the multi-scale sound field component, wherein the energy distribution characteristics comprise energy duty ratio and energy gravity center offset of each scale; Calculating the arrival time difference of the ultrasonic purification signal to obtain an acoustic wave propagation delay parameter; Performing sound source positioning calculation based on the sound wave propagation delay parameters to obtain a discharge source space coordinate positioning value; Combining the energy distribution characteristics and the discharge source space coordinate positioning values into sound source positioning characteristic vectors; and integrating sound source positioning feature vectors corresponding to all the ultrasonic purification signals into the ultrasonic feature vector set.
6. The method for monitoring partial discharge in real time based on an ultrahigh frequency signal and an ultrasonic signal according to claim 1, wherein when multi-modal information fusion is performed based on the ultrahigh frequency feature vector set and the ultrasonic feature vector set, generating a joint feature matrix comprises: Performing time dimension alignment on the ultrahigh frequency characteristic vector set and the ultrasonic characteristic vector set to generate a synchronous characteristic pair sequence; Calculating mutual information values of the ultrahigh frequency feature vector and the ultrasonic feature vector in each synchronous feature pair, and reserving synchronous feature pairs with the mutual information values larger than a reference threshold parameter to obtain a subset of strong correlation feature pairs; performing principal component dimension reduction on the ultrahigh frequency feature vectors in the strong correlation feature subsets to obtain ultrahigh frequency compression feature vectors; performing linear discriminant analysis on the ultrasonic characteristic vectors in the strong correlation characteristic subsets to obtain ultrasonic discriminant characteristic vectors; Performing feature cascading on the ultrahigh frequency compression feature vector and the ultrasonic wave discrimination feature vector to generate primary fusion features; And carrying out nonlinear mapping transformation on the primary fusion features, and mapping the primary fusion features to a high-dimensional joint feature space to obtain the joint feature matrix.
7. The method for monitoring partial discharge in real time based on an ultrahigh frequency signal and an ultrasonic signal according to claim 6, wherein when calculating mutual information values of an ultrahigh frequency feature vector and an ultrasonic feature vector in each synchronization feature pair, preserving synchronization feature pairs with the mutual information values larger than a reference threshold parameter, and obtaining a subset of strong correlation feature pairs, the method comprises: calculating a mutual information estimated value between the ultrahigh frequency characteristic vector and the ultrasonic characteristic vector by adopting an entropy estimation method based on K nearest neighbor; Normalizing all the mutual information estimated values, carrying out statistical distribution analysis, and constructing a mutual information value probability density distribution curve; determining a baseline threshold parameter based on robust statistical properties of the probability density distribution curve; And counting the number of the feature pairs in the candidate strong correlation feature pair set, and gradually reducing the reference threshold parameter according to a preset step length when the number of the feature pairs is smaller than a preset minimum pairing number until the number of the feature pairs meets the preset minimum pairing number, so as to obtain a final strong correlation feature pair subset.
8. The method for monitoring partial discharge in real time based on ultrahigh frequency signals and ultrasonic signals according to claim 6, wherein when performing nonlinear mapping transformation on the primary fusion features and mapping the primary fusion features to a high-dimensional joint feature space, obtaining the joint feature matrix comprises the following steps: inputting the primary fusion feature set into a kernel principal component analysis model, and calculating a covariance matrix of the primary fusion feature in a kernel feature space; And solving eigenvectors corresponding to the first M maximum eigenvalues of the covariance matrix, and arranging the eigenvectors in a descending order to construct the joint eigenvalue matrix.
9. The method for monitoring partial discharge in real time based on ultrahigh frequency signals and ultrasonic signals according to claim 1, wherein when constructing a partial discharge recognition model according to the joint feature matrix, the method comprises the following steps: determining a plurality of joint feature matrices based on the historical monitoring database; dividing samples in the joint feature matrix into a training data subset and a verification data subset; Performing iterative training on the deep neural network by using the training data subset until the loss function converges to a preset error range to obtain a trained model, wherein the deep neural network comprises an input layer, a hidden layer and an output layer, the number of nodes of the input layer is equal to the dimension of the joint feature matrix, and the number of nodes of the output layer is equal to the preset discharge type category number; And carrying out generalization capability assessment on the trained model based on the verification data subset, and reserving a model with assessment accuracy exceeding a preset performance threshold as the partial discharge recognition model.
10. The method for monitoring partial discharge in real time based on ultrahigh frequency signals and ultrasonic signals according to claim 1, wherein when obtaining the result of real-time diagnosis and monitoring of partial discharge of the tested power equipment, the method comprises the following steps: Inputting a new joint feature matrix determined in real time in the next stage into the partial discharge recognition model, and obtaining probability distribution sequences of various discharge types through forward propagation calculation; and selecting the category with the largest probability value in the probability distribution sequence as the real-time diagnosis and monitoring result of the partial discharge.

Description

Partial discharge real-time monitoring method based on ultrahigh frequency signal and ultrasonic signal Technical Field The invention relates to the technical field of discharge monitoring, in particular to a partial discharge real-time monitoring method based on an ultrahigh frequency signal and an ultrasonic signal. Background Partial discharge on-line monitoring refers to a technology for identifying early insulation defects and evaluating degradation trend of the early insulation defects by sensing the insulation state of power equipment in real time, and is widely applied to state maintenance of high-voltage power equipment. Partial discharge can produce a series of physical phenomena and chemical changes in the interior and surrounding space of the power equipment, such as light, sound, and the like. These various physical phenomena and chemical changes associated with partial discharges can provide a detection signal for monitoring the internal insulation state of the electrical equipment. The existing monitoring technology often relies on off-line detection and single-mode signal detection, which have obvious short plates, and cannot meet the requirements of a modern power grid, wherein the off-line detection is carried out in a laboratory or in a power failure state, the accuracy of detected data is limited unlike the voltage and load environment when equipment actually operates, the single-mode signal is easily influenced by electromagnetic interference or environmental noise, the identification accuracy is insufficient, and in addition, the single-signal feature dimension is limited, and the discharge essence is difficult to comprehensively describe. In addition, the prior art has multiple static threshold judgment, lacks continuous tracking and dynamic early warning capability on the evolution process of the insulating state, and is difficult to capture key turning points of accelerated degradation of defects in time. Disclosure of Invention The embodiment of the invention provides a partial discharge real-time monitoring method based on an ultrahigh frequency signal and an ultrasonic signal, which aims to solve the technical problems of low single-mode identification accuracy, insufficient multi-mode fusion and lack of state evolution prediction capability in the prior art, and realize the technical effects of dual-mode signal complementary enhancement, insulation state fine grading and trend evolution dynamic early warning, so that the early identification accuracy and predictive maintenance level of equipment insulation defects are obviously improved on the premise of ensuring the real-time diagnosis and monitoring accuracy of partial discharge. In order to achieve the above object, the present invention provides a method for monitoring partial discharge in real time based on an ultrahigh frequency signal and an ultrasonic signal, comprising: acquiring an original monitoring data set synchronously acquired when the tested power equipment operates, wherein the original monitoring data set comprises an ultrahigh frequency original signal sequence and an ultrasonic original signal sequence; performing signal purification treatment on the original monitoring data set to obtain a standard monitoring data set, wherein the standard monitoring data set comprises an ultrahigh frequency purification signal sequence and an ultrasonic purification signal sequence; extracting waveform characteristics of the ultrahigh frequency purification signal sequence, constructing an ultrahigh frequency characteristic vector set, extracting sound field characteristics of the ultrasonic purification signal sequence, and constructing an ultrasonic characteristic vector set; based on the ultrahigh frequency feature vector set and the ultrasonic feature vector set, carrying out multi-mode information fusion to generate a joint feature matrix; And constructing a partial discharge identification model according to the joint feature matrix to obtain a real-time diagnosis and monitoring result of the partial discharge of the tested power equipment. Further, when acquiring the original monitoring data set synchronously acquired during the running of the tested power equipment, the method comprises the following steps: controlling an ultrahigh frequency sensor probe and an ultrasonic sensor probe to acquire signals under the same time reference through a synchronous trigger mechanism to obtain an initial ultrahigh frequency signal segment and an initial ultrasonic signal segment; performing time stamp alignment processing on the initial ultrahigh frequency signal segment and the initial ultrasonic signal segment to generate a time synchronization signal pair; and integrating all time synchronization signal pairs in the acquisition period into the original monitoring data set. Further, when the signal purification processing is performed on the original monitoring data set to obtain a standard monitoring data set, the method in