Fault Classification of a Long Transmission Line using Nearest Neighbor Algorithm and Boolean Indicators

This paper proposes a plain-sailing yet powerful method of classification of 3 phase faults in a long transmission line using k-Nearest Neighbor algorithm approach and Boolean Indicators so as to correct the error in the former approach. Discrete Wavelet transform is carried out of the post fault current signal obtained from the model and standard deviations of the approximate coefficients of the each phase are obtained and accordingly a training matrix including the aforesaid feature along with its target and a sample matrix and its target are formed and classified using k-NN algorithm described in the paper. The error thus obtained is rectified using Boolean Indicators and comparing them with suitable thresholds.