Pattern recognition of partial discharge signal in gas insulated switchgear apparatus using visual and cluster analysis

This paper describes the development of an acoustic partial discharge (PD) detection system to discriminate the defect sources and the dimension of the particles in a compressed SF₆ gas insulated switchgear (GIS). Different types of material with different shaping are used as artificial defects to induce PD in experimental works. PD measurements have been obtained under AC stress in a uniform electric field configuration. Discrimination and classification of the PD distributions of the studied defects are possible using the technique of visual and cluster analysis. The aim of visual analysis is to display the discharge signals in 3 dimensional forms to increase the visible strength. Cluster analysis is used to study and calculate the dissimilarity for the statistical parameters for different kinds of discharge sources and present it in Dendogram.