Novel denoizing method for partial discharge signals using singular value decomposition and spectral subtraction

Partial discharge (PD) detection is essential in assessing the insulation state of electrical equipment. However, PD signals are often overwhelmed by interference, resulting in inaccurate detection results. Aiming at this problem, this study proposes a PD detection method based on singular value decomposition (SVD) and improved spectral subtraction. First, the test signal is constructed as a Hankel matrix, which is used as a trajectory matrix for the SVD. Next, the singular value mutation point in the feature matrix is set as the threshold for removing the narrowband interference (NBI), and a signal containing only white noise is obtained. Finally, the improved spectral subtraction is used to remove white noise and improve the signal-to-noise ratio (SNR). The method proposed herein, along with the variational mode decomposition, the empirical mode decomposition, and the improved threshold wavelet method, are applied to the processing of PD signals. Also, the SNR value, waveform similarity coefficient, and mean square error of the denoizing signal of the four algorithms were calculated, considering the noise suppression and feature preservation abilities. The simulation and measurement results show that the SVD-spectral subtraction method has a strong suppression effect on narrow-band interference and white noise. Compared with other algorithms, this method can significantly improve the execution efficiency and has great application prospects.