Detection ofHigh Impedance Faults inDistribution System

Thisinvestigation seeks topresent anewmethodof identifying highimpedance faults (HIFs)inthedistribution feeder. Discrete wavelet transformations (DWT)andneural networks (NN)havebeenwidely usedinpowersystemresearch. Consequently, this workdeveloped anovel methodtodistinguish effectively theHIFsbyintegrating DWT withNN.Theproposed schemehastwodistinct features. First, theinput signal ofthis algorithm isneutral linecurrent, rather thanthetraditional currentsbasedon threeindividual phases. Second, HIFs identification applies thedetails atlevels 2,3 and4 andthe approximations at level4 oftheneutral linecurrentare employed for. Theresults ofstaged fault clearly indicate thatthe proposed canaccurately findtheHIFsinthedistribution feeder. IndexTerms--arcing fault, discrete wavelettransform, downedconductor, highimpedance fault, neural networks