Wavelet De-Noising for PD UHF Signals Based on Adaptive Thresholding by Genetic Algorithm

Abstract

This paper presents an adaptive wavelet thresholding algorithm for de-noising of ultra-high-frequency (UHF) signals of partial discharges (PD). The wavelet de-nosing algorithm is based on an optimum and adaptive shrinkage scheme. A class of shrinkage functions with continuous derivatives and a genetic algorithm are used for the adaptive shrinkage scheme. The genetic algorithm is helpful to obtain global optimum thresholds and to reduce much time wasted by the adaptive searching computation. The de-noising results of PD UHF signals embedded in white noises are presented. The PD UHF signals denoised by the adaptive wavelet thresholding algorithm have smaller distortion in waveform than the signals de-noised by the soft thresholding algorithms.