Study On Aging Testing Of Generator Bar And Noise Elimination Of PD With Wavelet

The large generator is the key device of electrical power system,and its running reliability is very important to the whole system's stability,so it is highly concerned by people. It is very necessary to monitor the generator's insulation status that its insulation system is the main threaten of its safely running. Now the general insulation monitoring methods are PD (partial discharge) monitoring.In order to monitor the PD of generator's insulation,the valid characteristic quantity must be found that can reflect the PD variation and the relationship between the characteristic quantity and the insulation life. In this paper,the multifactors- electrical,thermal,cool and thermal circulation of accelerating ageing experiment based on the stator bar of 220MW turbine generator has been conducted in the laboratory. The results of the experiment show that there has not fixed rule between maximum PD quantity and insulation,but the increasing of maximum PD grads increase along with ageing time of insulation. So that the maximum PD grads is the characteristic quantity of evaluating the insulation condition of stator bar has been put forward. At the same time,the results show that the minimum PD quantity and its discharge numbers also add along with the aging time increasing. And the craggedness degree of - q figure of PD becomes steep along with ageing time too. So they also can be the characteristic quantity to evaluate the generator stator bar's insulation condition.The important section is denoising of PD during the generator insulation's PD monitoring. There have many kinds of noises in site,thereinto white noise and periodic narrowband noise are tow important noises. A practical approach has been brought forward that the priori information of signal and the external parameter don't needed,on the basis of summarizing the methods of white noise elimination with wavelet and the different Lipschtiz exponents and characters in time-scale plane of white noise and PD. This method can eliminate the white noise of PD,the tiny PD signal also can be preserved,the distortion is very little and it is suitable the site signalprocessing.Wavelet packet has better decomposition character than wavelet,and can more validly decompose the frequency space of signal. In this paper,eliminating the periodic narrowband noise with wavelet packet has been put forward. The periodic narrowband noise eliminating method has been proposed that has the optimal wavelet packet base. The instances show that this approach can eliminate the periodic noise of PD signal,and its dose not need set the external parameters.