Date of Publication :20th June 2019

Abstract: In Power system transmission lines are used for transmitting power to the load point. These transmission lines especially overhead transmission lines are susceptible to different types of faults – symmetrical and unsymmetrical faults. For ensuring the reliability of supply the determination of fault location is very important. Traditional methods for fault location were based on impedance and reactance methods. Both these methods are based on phasor quantities which produce delay in locating and detection of faults. Accuracy can be improved by processing travelling waves generated under fault condition. In the present paper these travelling waves are processed by using wavelet transform technique which gives the time and frequency response of the given wave. For achieving more accuracy Artificial Neural networks are trained under different fault conditions for locating faults. Wavelet transformation technique is used for extracting the three phase voltages and currents thereby avoiding additional filters. These quantities are fed as an input to the neural network and the output of this network will be the location of faults. Only difficulty with this method is the selection of required number of neurons for the network that gives accurate results. The simulations are performed using Artificial Neural Network toolbox in MATLAB

Reference :