A method for arc fault detection based on the analysis of signal's characteristic frequency band with wavelet transform

Abstract

Arc fault is one of main causes of fires which bring great loss and damage every year. It is identified that the traditional current breakers will not successfully interrupt the circuit all time, when an arc fault occurs. It is very important and necessary to find a new method to identify and interrupt the arc fault, although it is hard to be detected. This paper presents a method using wavelet transform to analyze current signal's characteristic frequency bands to detect arc faults. An experiment setup is designed to obtain both of the current in normal operation and the current with arc faults. Both of them are decomposed by wavelet function to obtain the characteristic frequency bands. The frequency bands of arc fault current represent a special character. By using this special character, the arc fault can be detected accurately and rapidly.