Research on fault diagnosis of operating mechanism for 12kV vacuum circuit breaker based on PSO-LSSVM

Abstract

The small fault samples and single characteristic parameters in vacuum circuit breakers (VCBs) would lower the accuracy and reliability of mechanical fault diagnosis. In this paper, the problem has been solved with applying a fault diagnosis method based on particle swarm optimization (PSO) and least square support vector machine (LSSVM). By analyzing the close coil current (CC) of VCB, the eigenvalues of time and current are extracted as input vectors. The paper uses PSO to optimize the model parameters of LSSVM, which are important to fault diagnosis, and to select the best subset of eigenvectors to obtain the optimal performance of LSSVM classifier. Then the improved support vector machine (SVM) is used to train and test the eigenvectors and different states of VCB are classified. Results show that the proposed method can detect whether VCB is normal or not. And the validity and accuracy is verified.