A pattern recognition application for distance protection

Abstract

In this paper, a method based on Support Vector Machines (SVM) as pattern classifier is presented for distance protection operation. The proposed method generates an output by determining faults occurred at different location of power transmission line for relay operation. The approach utilizes three-phase current and voltage information. The current and voltage information which are used in training and testing of SVM are obtained from a transmission line simulation which is performed by Alternative Transient Program (ATP). In the implemented digital relay application, it is shown that SVM classifies faults occurred on transmission line with very high performance level.