Classification of Fault Type on Loop-Configuration Transmission System Using Support Vector Machine

Abstract

This paper proposed to applied Support vector machine (SVM) algorithm for classified the fault type on the 500 kV transmission systems with connected in loop configuration. The fault signal was simulated using ATPDraw/EMTP program at frequency 200 kHz. The fault detection was analyzing the high frequency component by discrete wavelet transform (DWT). For the first stage, the coefficient of DWT was used for the fault detection. After the fault can be detected, the fault classification will be identified using SVM algorithm. The maximum coefficient from wavelet transform was used as input pattern of SVM to classify the type of fault. The input pattern of SVM consists of 4 input; maximum coefficient of DWT in all phase current and zero sequence current. For the SVM process, the fault classification used the five model of SVM because each model is working in parallel to avoid mistake (or error). In addition, the same input in five model were simultaneously used while the output of each models is differently according to specification of model. The overall result of 2160 case studies data can be summarized that the fault classification using SVM algorithm is highly satisfactory.