Circuit Breaker Fault Detection Based on CEEMD-GSA-SVM

To accurately extract the fault characteristics of circuit breakers, a fault detection method based on comprehensive sensitive empirical mode decomposition (CEEMD), gravitational search algorithm (GSA) and support vector machine (SVM) is proposed. Firstly, the CEEMD is used to process the circuit breaker signal, and using Hilbert transform to establish the marginal spectrum of the obtained components. Select energy entropy as the feature vector. Aiming at the problem that the parameter setting in SVM method affects the classification performance, using GSA to determine SVM parameters. Finally, Kernel-based Fuzzy C-Means (KFCM)- SVM is used for fault detection. In the end, the fault detection accuracy of the CEEMD-GSA-SVM method is 97.5%.