Understanding And Consideration On Interference Suppression Technology Of PD Signal Base On The Method Of Sparse Decomposition

Partial discharge(PD) is the import signals and manifestations of the power equipment internal insulation defects or insulation degradation. Effective partial discharge measurement for accurate assessment of power equipment insulation health has important practical significance. However, in the process of real-time online monitoring of power equipment, PD signal often influenced by the complex electromagnetic environment, so that there are a lot of complex noise signal in it. How to effectively put the useful PD signal and noise signal separation, it is the important and difficult in the research of power equipment on-line monitoring.This paper aiming at the shortage of current equipment PD signal interference suppression in power, adhibiting the idea of the sparse representation into the PD signal processing, and integrated the excellent performance of PD pulse signal extraction from wavelet basis function, in addition it used an idea which is the interference of white noise suppression from PD signals on the method of matching pursuit sparse decomposition based on wavelet cascade atom dictionary innovatively. The method of Matching pursuit(MP) can adaptively selected the best match of PD signal wavelet atoms from a over-complete wavelet cascade atoms dictionary, selected residual ratio as the terminating condition, and selected each iteration of the best match wavelet atoms represents a linear combination of PD signal.The feasibility and effectiveness of the method was validated by case studies in Exponential PD signals and UHF PD signals. With traditional global threshold wavelet decomposition, hierarchical threshold wavelet decomposition and wavelet packet decomposition method for comparative analysis, the results show that the method of matching pursuit sparse decomposition based on wavelet cascade atom dictionary in terms of PD signal and white noise interference suppression, has high accuracy, small waveform distortion, better retain the original advantages of the main characteristics of the PD signal information.For periodic narrowband interference suppression in PD signal, it is built the wavelet cascade atoms dictionary and Fourier atoms dictionary in this paper. First using Fourier atoms to extract PD signals the periodic narrowband interference ingeniously, it eliminates the periodic narrowband interference signal influence on PD signal interference suppression, improve the performance of the sparse decomposition of PD signal interference suppression. The feasibility and effectiveness of the method was validated by case studies in Exponential PD signals and UHF PD signals for the periodic narrowband interference suppression.And compared with the result of hierarchical threshold wavelet decomposition without periodic narrowband interference,it show that the method was superior in periodic narrowband interference suppression of PD signals.