ANN based power system fault classification

Abstract

This paper presents Wavelet based back propagation algorithm for classifying the power system faults, which is quite reliable, fast and computationally efficient. The proposed technique consists of a preprocessing unit based on discrete wavelet transform (DWT) in combination with an artificial neural network (ANN). The DWT acts as extractor of distinctive features in the input current signal which are collected at source end. The information is then fed into ANN for classifying faults. It can be used on-line following the operation of digital relays or off-line using the data stored in the digital recording apparatus. Extensive simulation studies carried out using MATLAB show that the proposed algorithm provides an accepted degree of accuracy in fault classification under different fault conditions.