FAULT IDENTIFICATION USING NEURAL NETWORKS

Abstract

In this thesis the aim is to detect the high impedance fault occurring on radial electrical distribution systems using neural network based relaying scheme. A multilayer perceptron is used for distinguishing the linear and nonlinear High Impedance Faults by taking the Feature vector as input. R.M.S values of third and fifth harmonic components of feeder voltage and feeder current are used as the feature vector obtained by applying the Fast Fourier Transform on the Feeder voltage and Feeder current.