A method of polarity extraction and type recognition of generator partial discharge

The polarity of large-scale generator partial discharge signal has importance to the partial discharge type recognition.Digital partial discharge online monitoring facilities are restricted by the strong disturb and the sampling frequency,so collected partial discharge signals have a lot of noises and lose some information about partial discharge.This brings big difficulty to polarity signal extraction.Wavelet transformation,which is time and frequency part changing through expansion and shrinkage and peace calculation,carries out much dimension refining analysis on the signal,and it is a good tool to remove signal noises.Firstly,we made up the frequency domain of the collected signal by FFT,and fixed the frequency range about partial discharge signal,and then took out the signal in the frequency range.Secondly,removed noises for extracted signal making use of Meyer wavelet that is very close to partial discharge impulse.At last,we counted the positive and negative polarity impulse in the fixed discharge voltage range for de-noised signal,and confirmed partial discharge type based on the relations between the partial discharge signal polarity and the discharge type.