A Novel Method of Selecting Complex Wavelet for Feature Extraction in Partial Discharge Signal Processing

This paper presents an efficient method for selecting optimal complex wavelet from a family of wavelets for feature extraction in partial discharge signal (PDS) processing. The optimal wavelet can be used to extract features of PDS from signals with strong disturbance and noise. Specifically, the optimal complex wavelet is selected based on the phase-spectrum similarity between the wavelet and the PDS. The simulation results show that the selected optimal wavelet cansignificantly improve the PDS feature extraction. This capability has potential to improve the accuracy of real-time monitoring of PDS in power systems.