Research on Partial Discharge De-Noising for Transformer Based on Synchro-Squeezed Continuous Wavelet Transform

Abstract

Traditional time-frequency methods for partial discharge (PD) de-noising have some limitations such as low time-frequency resolution, single de-noising type and poor readability. In this paper, a novel de-noising algorithm based on synchro-squeezed continuous wavelet transform (CWT) is adopted to filter out narrowband noise and white noise. The synchro-squeezed CWT algorithm is designed to redistribute the time-frequency domain and to distinguish the signal from the noise carefully as a high-rate time-frequency analysis. High-order statistics is employed to pre-process the polluted PD signal. The generalized cross-validation (GCV) threshold is combined with the adaptive trimmed threshold of synchro-squeezing (SS) domain to deal with the subsequent signals. The proposed algorithm can effectively suppress two kinds of noise, and the signal distortion is lower.