Study On The Loop Antenna For Partial Discharge Detection In Gas Insulated Switchgear And The Antijamming Technique With Complex Wavelet Tansformation

Partial discharge (PD) in GIS (Gas Insulated Switchgear) usually doesn't influence the safe operation of GIS, but long-time PD will deteriorate the insulation of the SF6 gas and result in insulation fault. In virtue of PD reflecting the status of GIS, detection of PD has been extensively researched and the key techniques are sensors, signal processing and pattern recognition. Therefore, this dissertation designed a ultra-high frequency shield resonance loop antenna (SRLA), by which lots of different PD signals have been sampled. Meanwhile. This paper introduces how to construct the complex wavelet applied to suppress interferences, and analyses the common and difference of the complex and real wavelet to suppress interferences. Finally, associated method of software and hardware is adopted to abstract PD signals. The main conclusions are shown as following:The theoretical characteristic parameters of SRLA are great close to the practical ones. Comparing the SRLA with the built-in loop sensor, conclusions are draw: SRLA has high sensitivity, good gain (1.95) and obvious directivity, and its SWR like "U" shape makes it well suppress outband interferences and be adapt to frequency selection detection.Paper introduces the idea and process of constructing complex wavelet according to the principle of constructing the wavelet, and presents the db complex wavelets. Four simulative PD signals are used to verify the accuracy of the complex wavelet transformation. Furthermore, it is proved that this construction is the same with the wavelet having scale function.Effective wavelet coefficient (EWC) threshold selection firstly put forward is applied to suppress white noise and narrowband interference. Comparing the EWC threshold selection with threshold selection of maximin theory and Stein unbiased risk estimate theory, it is found that EWC threshold selection is adaptive, and the effect of EWC threshold selection to restrain noise is better than that of the another two.Simulation and application of suppressing the interferences indicates that threshold selection and noise level greatly influence the effect of complex wavelet suppressing white noise, and the frequency of narrowband extremely affects the outcome of complex wavelet processing, whose amplitude's effect is alterative with its frequency. The ability of mulriple information distilling PD signal is much stronger than simpleinformation, and mulriple information WTRIPH is the best in this paper.