Analysis method of abnormal condition in GIS based on ultrasonic detection

Gas insulated switchgear (GIS) plays an important role in power systems. In order to monitor the online condition of GIS, the ultrasonic methods are widely used for detecting partial discharge (PD) owing to its anti-electromagnetic interference and accurate location. In this paper, time-frequency analysis and frequency spectrum analysis of ultrasonic original signals were carried out for extracting the features both in time and frequency domains. Based on plenty of experiment data and analyzing results, the features in time domain and phase distribution of typical PD such as floating metal discharge, corona discharge and free metal PD were discussed and summarized. The difference between PD signal and mechanical vibration signal was studied, it was found that pulse duration of mechanical vibration signal was longer, and its envelope was approximate to a sine wave, while the envelope of PD signal was nearly triangle-hump shape. It was obtained that PD signal had a wide spectrum of 20kHz~100kHz, meanwhile mechanical vibration signal had a narrower spectrum of 20kHz~50kHz. Ultrasonic associated with electromagnetic wave localization method, ultrasonic time delay localization method and ultrasonic wavelet time-frequency localization method were used for GIS on-site PD detection, all of them had good precision and accuracy.