Study on Filtering and Eigenvalue Extraction of Partial Discharge Signal of GIS

At present, GIS (Gas Insulated Switchgears) has been more and more widely used in ultra-high voltage power system. But its structure is complex to disassemble hardly, and difficult to locate and maintain in case of failure. Therefore, it is necessary to conduct partial discharge detection on GIS to master the types and characteristics of its internal defects and avoid further deterioration of insulation defects. This paper introduces the filtering processing of signals in detecting GIS partial discharge by uhf method. It extracts statistical operators as pattern discriminating characteristic values by drawing various spectral diagrams, and finally uses BP neural network to realize pattern recognition to judge fault types. A large number of experiments show that the method has a high recognition rate and a good experimental effect.