Application of wavelet transform to study partial discharge in XLPE sample

Partial discharge (PD) plays a paramount role in determining system reliability. But noise is always a major limitation of PD measurement. Many research work shows that wavelet transform (WT) has a potential to extract PD pulse from noise environment. De-noising algorithm based on WT has three steps: decomposition, thresholding and reconstruction. The selection of threshold and wavelet base is crucial to the whole procedure. Although some thresholding methods are effective in their own simulation environment, whether they can recover practical contaminated signals is uncertain. This paper chooses the most popular threshold algorithms to de-noise a contaminated PD signal that occurs in a cross-linked polyethylene (XLPE) sample and test their de-noising capability. Comparison of results by using different wavelet bases is also studied to point out the importance of proper selection of wavelet base.