An Adaptive Wavelet Multilevel Soft Threshold Algorithm for Denoising Partial Discharge Signals

For denoising of partial discharge (PD) signals in electrical equipment, the traditional wavelet threshold method is affected by the empirical and subjective selection methods of wavelet base and decomposition layer, which will cause the waveform to be distorted and produce large errors in the process of denoising. In order to reduce the influence of the above factors, this paper proposes a new adaptive wavelet multilevel soft threshold denoising algorithm based on waveform similarity parameters and energy distribution analysis. Firstly, the waveform similarity method is used to determine the optimal wavelet base. Secondly, multi-scale wavelet decomposition is performed on the simulated ideal partial discharge signal and noise, and the signal is processed by the multilevel threshold method. Finally, the energy analysis of the PD signal and the noise signal is performed to determine the optimal number of decomposition layers. The proposed method and the traditional wavelet threshold method are used to denoise the simulated discharge signal, and the denoising effects of the two algorithms are compared quantitatively from four indexes: signal-to-noise ratio, root mean square error, waveform similarity coefficient and ascending trend parameter. After calculation, the index of the new algorithm is improved, which indicates that the new algorithm has certain practicability and provides a new reference for the denoising of PD signals.