Noise Elimination And Characteristic Analysis Of The Partial Discharge Signals

Partial discharge is considered as one main reason that lead to electrical equipments damage, so partial discharge on-line monitoring is meaningful in precise interpretation of insulating condition. In this paper, partial discharge signals are discussed from several facets.With consider to different types of the noise possibly encountered in partial discharge signals, sometimes they submerged partial discharge signals, so de-noise partial discharge signals become a big problem at partial discharge monitoring. This paper is based on the wavelet transform combine with mathematical morphology method to eliminate the noises of partial discharge, and use self-adaptive methods to select the threshold coefficients. Wavelet transform threshold and wavelet transform modular maximum have compared with this method. These research results present useful design references for the application of the mathematical morphology-based filters in power signal monitoring and processing. For the strong narrow-band interference of partial discharge signals, a new method is put forward to suppress periodic narrow-band interference in partial discharge signals based on FFT and wavelet theory. Simulation proves that the wavelet transform combine with mathematical morphology method can be more effectively eliminate to the noises of partial discharge signals, and preserve the original signals very well. At the same time, the FFT and wavelet theory method can be eliminate the strong narrow-band interference of partial discharge signals, and the distortion of partial discharge signals is very small.This paper detailed analyzing the structure element of mathematical morphology, regarding the impacts of the shape, amplitude and width of the structure element of mathematical morphology-based digital filters.Characteristic analysis is used in internal discharges, slot discharges and end winding discharges. This paper analyzed the partial discharge signals from figure, frequency spectrum, energy spectrum and patter spectrum, and supplying a new idea for later research.