Research On Partial Discharge Signal Denoising Based On Wavelet Theory

Partial discharge is one of the most important reasons for the deterioration of electrical equipment insulation.Therefore,the on-line monitoring of partial discharge is of great significance for the timely and accurate understanding of the damage of electrical equipment in power system.The partial discharge of electrical equipment will produce a variety of electricity,light,sound and other phenomena,corresponding to the corresponding detection technology,Such as electrical detection and optical measurement,acoustic measurement and other non-power detection methods.At present,although there are a variety of detection methods,but most of them are specialized testing equipment,a variety of detection methods are independent of each other,it is difficult to horizontally compare.Therefore,the development of partial discharge detector is conducive to the operation and maintenance of substation equipment to meet the development requirements of power system.After the development of partial discharge combined detector,in the actual operation of the process,the scene has a variety of noise signals,which makes our partial discharge signal has been contaminated,and sometimes even completely detect the partial discharge signal,so the Partial discharge signal de-noising is very necessary.Based on the partial discharge detector,this paper discusses the application of threshold denoising method in partial discharge signal denoising,on the basis of studying the theory of partial discharge and wavelet.Aiming at the problems existing in the traditional threshold method,such as the selection of threshold function,the selection of threshold and the selection of wavelet bases,the paper makes a deep research.In the selection of the threshold function,the wavelet coefficients of the hard threshold function are discontinuous at the threshold,so that the reconstructed image has the pseudo Gibbs phenomenon.As for soft threshold method,although the processed signal is continuous,it will cause the image to have fixed deviation,which will affect the image denoising effect.Methods according to the problems above derived improvement,the improved threshold function is continuous and high order derivative,which has a very strong practical significance;In the selection of wavelet basis,the effect of different wavelet bases is different,and the better performance of db wavelet is determined by analysis;This paper also makes an in-depth discussion on the selection of threshold.Finally,the application of MATLAB toolbox by modeling and programming method of improved wavelet denoising theory is simulated.The simulation results show that the improved method can eliminate the noise of the partial discharge has a advantage of excellent mathematical properties.