Insulation Defect Diagnostic Method for OIP Bushing Based on Multiclass LS-SVM and Cuckoo Search

Abstract

Frequency-domain dielectric spectrum (FDS) is an effective testing method to reflect the changes of internal insulation status of oil-impregnated paper (OIP) bushing. In field application, the results of FDS test can be both affected by aging defects and damp defects, then the internal insulation defects of OIP bushing cannot be diagnosed in detail, which is an issue of discrimination of multiclass classification. To solve this problem, a method to diagnose the internal insulation defects of OIP bushing is proposed based on multiclass least square support vector machines (LS-SVMs) optimized by cuckoo search (CS) algorithm. First, the multiclass LS-SVM parameters are optimized by the CS algorithm. Then, the training data set is used to train the multiclass LS-SVM model, and the test data set is used for model testing. The experimental results show that the proposed method can effectively diagnose the internal insulation defects in detail, i.e., aging defects and damp defects. In addition, in the proposed method, the CS algorithm is better than the genetic algorithm and particle swarm optimization algorithm, and the convergence rate of the CS algorithm is faster than the other two algorithms.