Application of CL multi-wavelet de-noising in partial discharge detection

Abstract

In order to extract the weak partial discharge signals which have been submerged in the strong on-site background noise from power equipments, a new method is proposed based on the basic theory of CL multi-wavelet. In the first place, haar and balanced pre-processing method are used to pre-process the PD signals. And then, the vector thresholding method and the neighbouring coefficients method are used to de-noise. Lastly, the principle of translation-invariant multi-wavelet is applied to eliminate the Gibbs phenomenon. Besides, this paper compares the effect of the two de-noising methods based on different noise. After analysis and comparison of the results, the conclusion is obtained that the neighbouring coefficients method can achieve better de-noising effect. Finally, the method is applied to de-noise the test data of generator partial discharge. The results show that it has the ability to retain the original signal information.