Comparison of Wavelet, Stationary Wavelet and Wavelet Packet Methods for De-noising of Partial Discharge on Power Cable System

摘要

The failure of Power Cable systems is mainly attributed to Partial Discharge activity in power Cable insulation and accessories such as terminations and joints. Different types of sensors are available for online and offline monitoring of Partial discharge activity. However the main difficulty involved in the Partial discharge measurement of Power Cable system at site is to eliminate the interference from other external and internal high frequency noisy signals. In this paper, de noising the PD signals by the wavelet transforms, stationary wavelet and wavelet packets techniques are compared for its merits and demerits. The method used for selection of mother wavelet is based on minimum entropy criteria. The two types of mother wavelet debauchees and symlet are used. The effectiveness of level dependent hard thresholding method in retrieving the original partial discharge signals over the other conventional thresholding methods is also analyzed. The laboratory measured Partial discharge signals are mixed with white Gaussian noise and discrete spectral interference (DSI) signals in such a way that the partial discharge signals are completely immersed in the simulated noises. The partial discharge signals are measured for power cables system with Improper Stress control tubing termination.