Suppressing the discrete spectral interference of the partial discharge signal based on bivariate empirical mode decomposition

Abstract

Summary In the present study, a discrete spectral interference (DSI) suppressing method based on bivariate empirical decomposition mode (BEMD) related to partial discharge measurements is presented. PD complex-valued signals are generated first, whose real part is DSI polluting PD signals and imaginary part is zero mean white Gaussian noise (WGN) signals. PD complex-valued signals are decomposed based on BEMD, and intrinsic mode functions (IMFs) can be sifted by this decomposition. The real part of IMFs (RIMFs) are decomposition results of DSI polluting signals and imaginary part of IMFs (IIMFs) are decomposition results of zero mean WGN signals. With the assistance of WGN signals, the mode mixing effect in RIMFs can be alleviated. Besides, RIMFs are free from the influence of WGN signals, which will avoid the disadvantage of long-time computing for ensemble empirical mode decomposition (EEMD) and satisfy the request of PD online monitoring. The goal for suppressing DSI of PD signals can be achieved by reconstructing RIMFs based on sine amplitude judgment method proposed in this study. The DSI suppressing method presented in this article was applied on the simulated and measured PD signals, and the obtained denoising results mainly based on EEMD, BEMD, and wavelet transform (WT) were compared and analyzed to verify the effectiveness of the proposed method. Besides, assessment indices including signal to noise ratio (SNR), amplitude relative error (ARE), root mean square error (RMSE), and normalized correlation coefficient (NCC) were used as objective evaluation criteria for denoising effect, and the results show that the presented method is superior to traditional EEMD methods and WT methods, which has less amplitude error as well as less waveform destination.