Two-stage Partial Discharge Location Method Based on Defect Recognition Pre-screening

Abstract

Partial Discharge (PD) detection of gas insulated substation (GIS) must be done before commissioning, this is an important guarantee to reduce the incidence of post-commissioning faults. Fast and accurate partial discharge location is the key to improve test efficiency. Therefore, a two-stage partial discharge location method based on pre-screening of defect recognition is proposed in this paper. The convolution neural network (CNN) is used to supervise and train the signal atlas recorded by GIS withstand voltage test. By identifying the discharging atlas, the types of defects can be distinguished. Based on analysis of the typical discharging defects in GIS and their affiliated area, a law of partial discharge detection area screening is established which can be used to exclude some gas chambers from consideration. Ultra High Frequency Time Difference of Arrival (UHF TDOA) positioning method is then used to carry out further positioning analysis to determine the accurate position of PD. The analysis shows that, combined with the pre-screening of defect recognition in GIS, the location judgment area can be effectively reduced and the complexity of positioning can be lowered.