Faults Classification for Transmission Line using Wavelet and Radial Base Function

Abstract

This paper presents a framework for fault classification using discrete wavelet transform (DWT) to prepare the feature which would be given as the input to the radial basis function neural network (RBFNN) which has high performance level especially in classification problems. The simulated network carried out using MATLAB/SIMULINK software package for three phase transmission lines which including the series compensator as it is very challenging task in line protection and other online decisions using different ten fault types, different locations along the transmission line and with different fault inception angle.
Discrete wavelet energy calculated from quarter cycle of only sending post fault current signals side is used as an input to single RBFNN which is trained and tested to provide all fault types. The method responses very fast with few numbers of training samples and can also detect the ground fault cases as good as other phases without any additional data.