Selection of entropy based mother-wavelet and level dependent threshold techniques for de-noising of partial discharge signals

Summary

Partial discharge measurement is considered to be one of the most important nondestructive diagnostic tool for condition monitoring of insulation systems of High Voltage equipment. However the main difficulty involved in the Partial discharge measurement of HV equipment at site is to eliminate the interference from other external and internal high frequency noisy signals. The two broad types of interferences that pollute the PD signals are random internal noise and externally coupled disturbances. Wavelet transforms have become popular in many areas of damage detection and feature selection due to their proven advantages, such as adaptability and high-resolution. The simultaneous frequencyand time-domain analysis available with wavelet transform makes it a powerful tool in extracting the PD signal from severe noise and interferences. Wavelet based de-noising can be performed using various techniques and each has its own advantages. In this paper, denoising by the wavelet transform and wavelet packets technique are compared and suitable method for selection of mother wavelet based up on entropy calculation, decomposition, & threshold methods are derived and compared on simulated and real signals. From the study it is observed that Mother wavelet selection based on minimum entropy is superior to that of Maximum Energy & Maximum Ratio selection. It is also observed that for all kinds of signals including severely corrupted signals, the level dependent thresholding is able to reproduce the original signal from the noisy signal than that of other conventional thresholding methods. For Real or field signals hard thresholding is giving better results than that of Soft thresholding. Wavelet packet analysis is also equally good as that of wavelet analysis with level dependent thresholding in retrieving the original signals with both simulated and real signals. In case of real signals, wavelet packet is able to produce low PAD (Peak Amplitude Distortion) compared to that of wavelet analysis.

Additional informations

Authors

K.P. MEENA, N.R. BURJUPATI

Keywords

wavelet techniques, mother wavelet, denoising, partial discharge, thresholding