Comparative Analysis of Intelligent Techniques for Fault Detection Classification and Location of Power System on Transmission Lines

Abstract

ABSTRACT newlineA system for transmitting power or even for distributing it suffers from the fault newlinewhich inevitable due to unforeseen events. What can be done here is that a system newlinecould be developed smart enough to detect this fault condition as soon as possible. newlineThen it should be able to rapidly to find the type of it i.e. the process of newlineclassification and at last quickly points the location of fault. newlineThe work presented in this theses addresses this problem effectively with the help newlineof Artificial Neural Network (ANN) alongside the discrete wavelet transform newline(DWT). Based on these techniques a fault locator (FL) and fault classifier newlineAlgorithm has been developed here. The proposed FL aims to detect and classify newlinefaults on the network with 3 phases for the process of transmitting power. It was newlinetrained from set of data gathered after wavelet analysis to achieve the aim as newlinementioned. The analysis has considered either ends (relay and far end). newlineFor developing the network to achieve the given task, both the algorithm of ANN newlineFeed forward and Back Propagation have been utilized here. Basically the newlinenetwork consists of three distinct layers out which first layers is the input layer newlineand last layer is the output layer. The second layer is called as hidden layer and it newlinehas two layers in it. newlineAlso, a comparison between the for fault classifier developed here by the use of newlineANN (ANNFC) and the Algorithm based Fault classifier (AFC) have shown the newlineresults going in the favor of the scheme proposed here. Since, the proposed newlineANNFC have nearly 100 % accuracy while AFC has 98% accuracy for all the newlineknown faults. newline