Noise suppression in PD signal based on Prony time series energy spectrum

Abstract

On-line partial discharge (PD) monitoring is related to the safe and stable operation of power grid. However, the measured signal is inevitably affected by noise. Among a variety of noises, white noise is the most difficult to suppress and ubiquitous. It mainly determines the Cramer-Rao lower bound that cannot be crossed by all methods. This article proposes the Prony time-series energy spectrum method to suppress white noise in a low signal-tonoise ratio environment. This method is divided into two parts: First, Prony time series energy spectrum is used to detect whether there is PD signals in the measured signal. Then, based on the least square method, the PD signal is estimated with the Prony model. In data analysis, this method can suppress -10dB white noise, while other methods have performed poorly or even failed under the same conditions. In general, this method has excellent anti-noise performance compared to several other algorithms.

Partial discharge signal compression reconstruction method based on transfer sparse representation and dual residual ratio threshold

“…Currently, the CS theory research on PD signals focuses on denoising, pattern recognition, and discharge localization. In [17], an over‐complete wavelet dictionary PD signal denoising algorithm based on the CS theory is proposed. A primal‐dual interior point and basis pursuit algorithm are proposed to realize the classification of PD signals of different fault samples [18].…”