Application of DWT and ANN for fault classification and location in a series compensated transmission line

In this paper an accurate approach is proposed for classifying and locating faults in series compensated (SC) network utilizing discrete wavelet transform (DWT) and artificial neural network (ANN). In the first phase of the proposed approach, fault current samples acquired from simulation were decomposed using Db5 mother wavelet. Faults signatures are captured in terms of standard deviation of detail coefficients of 1
^st
and 5
^th
level, norm entropy value of the wavelet coefficients and minimum and maximum value of the wavelet coefficients. In final phase, captured features are applied as input to classifier and distance estimator models for training and testing. For investigating the efficacy and accuracy of the proposed two-stage approach of classifying and locating faults in SC network, numerous test simulation have been carried out on a 400 KV, 300 km SC network in MATLAB environment. The results obtained from test cases, reveals that the proposed approach can reliably classify and locate the faults with high accuracy.