Research on wavelet-based sparse representation of insulator leakage current signal

Abstract

Summary In recent years, there are more and more frequent and severe accidents caused by pollution flashover on the substation equipment of Chinese power systems. This paper presents a method for processing the current leakage signal for pollution flashover, which is based on sparse representation, because the leakage current as a detection factor carries large amount of information and thus, requires higher sampling rate and processing speed. Firstly, under a lower sample frequency, sparse decomposition of leakage current signals for the pollution flashover of 110 kV transmission line is finished based on sym8 wavelet. Then, the signals are reconstructed by the Regularized Orthogonal Matching Pursuit algorithm. The result shows that sparse and quick representation of the signal is achieved. As the leakage current signal contains more data and more singularities, the reconstruction error is lower with relatively higher compression ratio. We argue that this sparse representation method will be put into broad applications for fault monitoring and signal transmission of power system. Copyright © 2013 John Wiley & Sons, Ltd.