Cable ground fault diagnosis based on current waveform by using support vector machine

Abstract

Recently the transmission and distribution department have time cost, human cost and money cost because of Electric power system innovation. So, when a distribution line accident occurs, it is necessary to automate the cause discrimination. In this research we did cable ground fault diagnosis based on current waveform by using support vector machine. We did a discrimination experiment in 2 steps by using support vector machine, and evaluation learning performance by using leave-one-out cross validation. As a result, we obtained high discrimination rates for both steps and both 10kHz and 100kHz. From this research, we were able to diagnose cable ground fault with a high discrimination rate by using support vector machine.