Identification of conductive leakage signal in power cable based on multi-classification PSO-SVM

Abstract

In order to effectively identify and extract the red signal from the leakage signal in the computer's power line, a method based on particle swarm optimization (PSO) is proposed to optimize the multi classification support vector machine (SVM). Firstly, the conductive leakage signal is filtered. Then the SVM penalty parameter and kernel parameter are optimized by PSO, the conduction leakage signal is trained and classified. Finally, the classification performance of the un-optimized SVM is compared with PSO-SVM. The result shows that this method has higher classification rate than the grid search method. The classification and recognition rate reaches 87.23%.