Method for identifying discharge types of partial discharge signals of electric power equipment

Abstract

The invention discloses a method for identifying discharge types of partial discharge signals of electric power equipment. The method comprises the steps of: S1, acquiring a partial discharge time domain waveform of a known type, and establishing a partial discharge type fingerprint database; S2, using a wavelet packet to perform denoising preprocessing on the partial discharge signals; S3, performing S transform on the partial discharge signals to obtain a time-frequency matrix A; S4, performing spectrum kurtosis decomposition on the partial discharge signals, and extracting a time-frequencymatrix B; S5, conducting singular value decomposition on the time-frequency matrix B, selecting an appropriate number of singular values, and normalizing the singular value as an eigenvector; S6, regarding an eigenvector of a training sample as an input of a probabilistic neural network optimized by adopting a genetic algorithm, and training to obtain the optimal probabilistic neural network structure; S7, and inputting a test sample eigenvector into the optimal probability neural network structure to obtain a partial discharge signal type. The method disclosed by the invention has the advantages of high identification speed and high identification precision.

Description

technical field [0001] The invention relates to the field of electric power technology, in particular to a method for identifying discharge types of partial discharge signals of electric equipment. Background technique [0002] Partial discharge of power equipment is not only a manifestation of equipment insulation defects, but also the main cause of equipment insulation damage. Therefore, partial discharge detection of equipment is an important means to find hidden faults in time and ensure safe and reliable operation of equipment. Partial discharge signals from different sources cause different harm to equipment, and the current partial discharge on-line monitoring system has a weak ability to identify the type of partial discharge signals, which has caused great trouble to the maintenance personnel of operation and maintenance personnel. Therefore, the study of partial discharge characteristics The extraction and discharge type identification algorithm is of great signifi...

AI Technical Summary

Problems solved by technology

Partial discharge signals from different sources cause different harm to equipment, and the current partial discharge on-line monitoring system has a weak ability to identify the type of partial discharge signals, which has caused great trouble to the maintenance personnel of operation and maintenance personnel. Therefore, the study of partial discharge characteristics The extraction and discharge type identification algorithm is of great significance to the safe operation of the power grid

[0003] In the prior art, the wavelet transform is generally used to extract the characteristic parameters, and the multi-layer perceptron based on the BP neural network is used to identify the partial discharge signal. Although compared with the traditional φ-q-n parameters, the dimension of the signal feature vector is reduced and the recognition is improved. Accuracy, but the wavelet transform is easily disturbed by noise, and sometimes the analysis conclusions at multiple scales will be contradictory, and the BP neural network is not a special algorithm for classification, and the classification results are mainly represented by rounding, which lacks intuition. In terms of algorithm optimization, too many weight and threshold parameters need to be optimized, and good classification results may not be obtained. In terms of network construction, because the BP neural network needs to be continuously debugged to obtain the best BP network structure, the steps are cumbersome. time consuming

Images