Medium voltage cable partial discharge signals denoising by Second Generation Wavelet Packet Transform

On-line partial discharge (PD) measurements taken from distribution cable network are corrupted by noise due to electromagnetic interference. This paper presents application of Second Generation Wavelet Packet Transform (SGWPT), as an improved algorithm of Second Generation Wavelet Transform (SGWT), to extraction of PD pulse from background noise. Firstly, the SGWPT is constructed on the basis of SGWT. Secondly, comparative study of the Shannon entropy and threshold entropy criterion to select the best tree is conducted, and the investigation results show that threshold entropy criterion yields better signal-to-noise ratio (SNR). Finally, the algorithm is applied to PD signals corrupted by white noise. Results confirm that SGWPT can improve the effectiveness of PD de-noising.