Cross-linked cable partial discharge mode recognition method based on parameter optimization SVM (Support Vector Machine) algorithm

Abstract

The invention discloses a cross-linked cable partial discharge mode recognition method based on a parameter optimization SVM (Support Vector Machine) algorithm. In order to avoid problems of a low neural network convergence rate, over learning and the like or low recognition precision due to improper SVM parameter section in traditional mode recognition, while an M-ary classification theory is introduced to extend the SVM algorithm with stronger generalization and learning abilities into multiple classifications of classifiers, an improved genetic algorithm is used for optimizing a penalty factor of each sub classifier and kernel function parameters, and thus the optimal parameter SVM classification model is built. Results show that each defect recognition rate is more than 95% when the optimal SVM serves as the classifier, and whether to optimize the parameters, the SVM general recognition ability is better than that of an RBF neural network.

Description

technical field [0001] The invention relates to the technical field of cable partial discharge pattern recognition, in particular to a cross-linked cable partial discharge pattern recognition method based on a parameter optimization SVM algorithm. Background technique [0002] In recent years, cross-linked polyethylene (XLPE) cables have gradually become the mainstream equipment for power transmission in my country's distribution network due to their reasonable process structure, simple installation and laying, and no oil leakage. Reliability is closely related to grid stability. In addition to external factors, partial discharge (hereinafter referred to as partial discharge) is the main cause of cable insulation failure. The partial discharge signal collected during the partial discharge detection process of the cable carries all the insulation fault information of the cable. The characteristics of the partial discharge signal generated by different defects are different, a...

AI Technical Summary

Problems solved by technology

[0004] (1) directly to various partial discharge The method of analyzing and comparing the three-dimensional spectrogram and its waveform and frequency spectrum, although this method summarizes the identification rules with certain reference value, but this identification method largely depends on engineering experience, and the objectivity is not strong

[0005] (2) Directly use the PD pulse time-domain waveform data value as the method of pattern recognition discharge fingerprint. Although this method simplifies the feature extraction process, even after dimension reduction processing, the recognition process may still encounter "dimension disaster".

[0006] (3) The method of extracting the fractal dimension of the partial discharge grayscale image as the input of the neural network, the feature dimension of this method is moderate, and a good recognition effect has been achieved, but the artificial neural network method lacks the support of mathematical theory and has a slow convergence speed And it is easy to fall into the disadvantage of local minimum value. Improper selection of network type and parameters will have a great impact on the classification results. It is a nonlinear classification algorithm that depends on experience.

[0007] (4) Using the cable partial discharge mode method of Support Vector Machine (SVM) based on statistical characteristics of partial discharge, although it can achieve good recognition results, the parameters of the SVM used are mostly dependent on manual debugging, and it does not take into account if the SVM Improper selection of parameters will have a negative impact on the accuracy of pattern recognition and the running speed of the algorithm, and the speed and accuracy of recognition cannot be guaranteed

Images