Experiences with data driven UHF PD rapid location detection for gas insulated switchgear

Gas insulated switchgear (GIS) plays a vital role in substation. Partial discharge (PD) is the most common and serious defects, and it greatly influences the insulation quality of GIS. So it is very important to locate the defect in time. The current methods of location are mainly based on the sensors time difference. The method is complex and it makes an inadequate use of monitoring sequence. So experiences with data driven UHF PD rapid location detection for GIS are shared in this paper. The experiences use monitoring sequences of different sensors to identify internal PD and locate the PD source. It includes two parts: identification of internal discharge, and rapid location of PD region. Firstly, sensors are grouped by the electrical correlation and space distance. And the correlation coefficients of different grouped sensors monitoring sequences are compared. Discharge types are identified based on the analysis of the correlation. Secondly, grid partition location method based on discharge sorting is used to estimate the PD source region for the internal discharge. Considering the UHF discharge signal decreases with the increase of propagation distance, the nearest sensor detects the maximum signal. The signals of electrical adjacent sensors are ranked, and sensors space grid are build. According to the GIS topology and the sensors spatial position information, the space around the sensor which receives the largest signal is divided into several regions. And the discharge is located where is near to the sensor which receives the second large signal. This method is proved to be feasible in locating discharge position, and reduces the workload and work time of maintainers.