Signal extraction methods for online partial discharge (PD) measurement of power transformers

Wavelet transform-based Partial discharge (PD) signal extraction methods have been widely adopted in the past ten years. However, these methods may still encounter some difficulties in online PD measurements, in which PD signals can be overwhelmed by noise. To overcome the limitations of wavelet transform-based methods, the authors of this paper have developed two novel signal extraction methods. One is a multiscale thresholding based wavelet transform, which adopts multiscale thresholds and provides probability indices for extracted signals to indicate their likelihood of being PD signals. Another is a differential PD signal extraction method, which is based on evaluation of changing rates of acquired signals. It can be applied to online PD measurements even when multiple PD sources occur simultaneously in a transformer. This paper demonstrates the applicability of the two methods on the extraction of PD signals from online PD measurements of field transformers. Comparisons of the two methods are also provided in this paper.