Partial Discharge Signal Extraction from Different Kinds of Noise based on Second Wavelet Transform

摘要

Wavelet transform has the characteristics of multiresolution analysis, and it has time-frequency localization characteristics, which is suitable for smooth, the non-stationary signal analysis. There are different kinds of noise in field collection of partial discharge signals and partial discharge is so weak that it is almost submerged by the noise. In paper, wavelet transform is used to de-noise different kinds of noise, including rand noise, periodic narrow-band interference, and both the two noise hybrid in partial discharge. Partial discharge simulation model are exponential damping or exponential damping oscillating. Considering original wavelet transform, that is first Wavelet transform has spectrum mixed, and global threshold de-noising effects is not ideal. So, automated threshold algorithm of second wavelet transform is used. Simulation results show that second wavelet transform auto threshold is effective in extract partial discharge from different kinds of noise, and it is adaptive to dispose on line.