Entropy-based wavelet de-noising for partial discharge measurement application

One of the major challenges in on-site Partial Discharge (PD) measurement is de-noising PD signals, which are normally being coupled with strong external noise. Therefore, Wavelet Transform (WT) techniques are being adopted in PD signal extraction. However due to their inherent shortcomings, online PD measurement may give wrong assessment. This paper proposes an Entropy-Based Wavelet Transform (EBWT) de-noising scheme for PD measurement using entropy distributions to extract real PD signals from noise-corrupt PD signal. This method chooses the best decomposition level first and then de-noised PD signals by selecting optimum wavelet base using EBWT and also compare the results with other methods. Finally the performance of proposed technique verified through case study on data acquired from PD measurements on experimental PD model.