For improving the adaptability of white noise de-noising algorithm

this paper presents an approach of choosing wavelet

level and threshold adaptively for de-noising of partial discharge（PD）signals. Firstly

on the basis of layer-by-layer wavelet decomposition of signals

energy percentage of approximate coefficients in all coefficients and wavelet energy entropy are calculated to represent the energy characteristics and shape characteristics. And the wavelet with maximum ratio will be selected. The energy and shape characteristics are taken into account simultaneously. Secondly

the algorithm applies white noise detection principle on detail coefficients of each level to get adaptive decomposition level. The decomposition will be terminated when the detail coefficients cannot meet white noise standards and the error will be reduced. Finally

the algorithm uses the whale optimization algorithm to optimise the process of getting adaptive threshold

improving the accuracy and speed. For verifying the de-noising results

PSO algorithm and maximum signal energy algorithm are used for de-nosing simulative PD signals and real PD signal. The results show that the algorithm proposed in this paper has the lest MSE and amplitude error and has a promising prospect in practical application.