Study Of On-line Detecting Partial Discharge Technology For 1000kV GIS Based On Multi-sensor

Gas Insulated Switchgear is the combined switch electric appliance with metal shell which enclosed the circuit breaker, isolating switch, grounding switch, bus and other equipment with SF6 gas(as insulation and arc extinguishing medium), with advantages of small footprint, safety coefficient high, widely used in EHV and UHV transmission and transformation project. 1000 kV GIS is the key device of UHV power transmission project, once faults occur, may cause the occurrence of major power accident. Insulation lower is a major cause of GIS equipment failure, on-line detection with Partial Discharge of GIS equipment(Partial Discharge, PD) can effectively control equipment insulation status, to prevent the GIS equipment fault from tripping which caused by power grid accidents. Recently, the research of PD on-line detection method in the 1000 kV GIS Equipments is based on single type sensor, may often appear the false alarm omission phenomenon in engineering practice. The research of PD multi-sensor on-line detection technology with the 1000 kV GIS equipments, has important practical significance to safe operation of power grids.This paper researches how to solve the problem of on-line detection device misstatement omission. First studies the generation mechanism and route transmission of PD and common GIS insulation fault, analyses the principle, characteristics and existing problems of three detection methods: ultrasound, UHF and SF6 gas decomposition component. Then the experimental platform is built to simulate several typical insulation faults, it concludes the reason of misstatement omission through analyzing detection results of three on-line detection methods. Finally, according to the characteristic of the ultrasonic sensor, UHF sensor and gas sensor three different principle sensors can receive the different types of PD signals, it puts forward on-line detection method of three different types of sensors information fusion. Multi-sensor information fusion online detection system is designed, using BP(Back Propagation) neural network and D-S evidence theory to fuse and decide the data acquired by three different types of sensors. Through detecting two kinds of typical faults using this system, it concludes that the system has high accuracy and reliability.Information fusion detection system uses BP neural network and D-S evidence theory to fuse location information of PD, adopts D-S evidence theory to do decision-level fusion of three single method fault types recognition results, realizes accurate location of PD faults and precise identification of PD types, solves the problems in actual application based on a single type sensor detection device, has the reference value of researching insulation condition detection of 1000 kV GIS equipments.