Skip to Main Content
Condition monitoring for gas insulated switchgear (GIS) requires an accurate and reliable identification of the defective equipment in it for maintenance purposes. In this paper, a feature extraction procedure is explored, which is based on the power spectral density (PSD) of the denoised partial discharges (PDs) emanating from the defective equipment in the GIS. Furthermore, artificial intelligence techniques, in particular, the self organising map (SOM), are investigated for their roles as classifiers to precisely identify this defective equipment, based on the PSD feature patterns. The performance of the SOM-based classifier is ascertained by using the PDs acquired from GIS in the Korean 154-kV EHV transmission networks.