A new technique for extracting partial discharge signals in on-line monitoring with wavelet analysis

This paper deals with the multiresolution analysis of wavelet transform for on-line partial discharge (PD) monitoring. One of the problems of PD on-line monitoring is how to suppress strong noises such as narrow-band radio frequency noise and random noise. In recent years, wavelet transform has become a powerful tool to analyze and process signals in various science and technology fields. In this paper, Mallat´s algorithm of dyadic wavelet transform is adopted for the research of the characteristics of PD signals and noises, and a new denoising technique with wavelet analysis is proposed. Traditionally, noise is filtered by using low-pass, bandpass or high-pass filter. While the new denoising technique carries out filtering on the base of the different singularity of PD signals and noises. The numerical simulating signals and some experimental data obtained in a power plant were processed by using the new technique. The results show that the new technique not only can improve the Signal to Noise Ratio (SNR), but also possesses high time resolution of PD signals. Finally some factors influencing the efficiency of filtering are discussed in detail