Fault Diagnosis Method of Circuit Breaker Based on CEEMDAN and PSO‐GSA‐SVM

Abstract

Due to that the feature extraction of vibration signal to judge the mechanical fault process of circuit breaker affected by interference, a fault feature extraction method based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and support vector machine (SVM) is proposed. Several intrinsic modal function (IMF) components reflecting the mechanical state information of the circuit breaker during operation are extracted from the vibration signal through CEEMDAN. According to the correlation coefficient and energy distribution of each IMF, the first seven IMF components are denoised by wavelet packet soft threshold, and the sample entropy is calculated as the feature quantity. Finally, the improved gravitational search algorithm (GSA) based on particle swarm optimization (PSO) is used to optimize the SVM classifier to classify and identify the different operating states of the circuit breaker. The experimental results show that the entropy characteristics of CEEMDAN samples are insensitive to signal interference, and the PSO‐GSA‐SVM diagnosis method has a good effect on fault identification in the opening operation process of high voltage circuit breakers. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.