Feature Extraction ofHighImpedance Arcing Faults inCompensated MV Networks. PartII:DWT-BasedAnalysis ofResidual Components

Electrical faults duetoleaning trees havebeenstudied in(1)-(2) anditwasfoundthattheinitial transients duetothe associated arcreignitions arebehavioral traits. Thesetransients havebeenlocalized using thediscrete wavelet transform (DWT). Inthis paper, these features areextracted fromresidual voltages using DWT.Therefore, thefault localization iscarried out.The termforlocating thefaulty feeder isbasedonthepolarity ofa specific frequency band-power computed bymultiplying theDWT detail coefficient oftheresidual voltage andcurrent foreach feeder. Theperformance oftheproposed technique isgetting better whentheresidual waveforms areutilized. Test cases provide evidence oftheefficacy oftheproposed technique.