Research On Three-phase Cylindrical Gis Partial Discharge Detection Method Based On UHF

Although there has been a great success on the research of partial discharge, the currentstudies are basically concentrated in coaxial structure GIS, few of them are concerning aboutthree-phase barrel GIS which is widely used across the world. The three-phase barrel GIS ismore compact in structure compared with three-phase separated barrel GIS, they can barehigher electrical voltage inside and have greater requirements on the insulation performanceof the equipment. Therefore, it is highly necessary to specially carry on research about partialdischarge in three-phase barrel GIS.After reading related literature, various GIS partial discharge detection methods aresummarized and analyzed, and the Ultra high frequency (UHF) method is selected as thepartial discharge detection method of this paper. The partial discharge in GIS can generatehigh frequency electromagnetic wave, its propagation and attenuation characteristics as wellas UHF detection methods are specially analyzed to rationally place the UHF sensor.The simulation software of ANSOFT12is used to simulate and calculate thethree-dimensional electrostatic field inside the three-phase barrel GIS both with and withoutdefect case. the internal electrostatic field distribution law is studied and the initial dischargefield strength is determined, laying a theoretical basis for discharge detection experimentalmodel and defect case model.The partial discharge test platform is built. The platform is composed of three parts,which are testing equipment, discharge detection facility and discharge model. Three partialdischarge patterns generated by different forms metal particles in three-phase barrel GIS aretested and researched. The three types of metal particles are fixed spikes on high voltageconductor, fixed spikes on inner wall and fixed metal particles on disc insulator. Thespectrums of discharge are analyzed to build the spectrum feature library, which is a referencefor partial discharge on-site detection. Five statistical features of degree of offset, thesteepness of the local peak points, cross-correlation coefficient and the discharge factors areextracted. the defect size’s influence on the initial field strength is studied, which has animportant value on partial discharge recognition.