Separation of Partial Discharge Faults in Metal-clad Switchgear Based on Pulse Shape Analysis

Switchgear is an integral part of power systems network. Any fault occurring in switchgear may leads to significantly long outage duration and the affected load, or area, can be enormously large. Partial discharge (PD) is one of such faults which are very common in metal-clad switchgear due to insulation defects in bushings, CTs, PTs, bus bar insulations and connections. Online and offline PD measurements have been recently employed to investigate the health condition of such insulations. It is more likely in such equipment that multiple PD faults exist concurrently and hence the identification and separation of multiple faults sometimes gets difficult. This paper presents separation of multiple PD faults in metal-clad switchgear based on the significant features extracted from PD pulse shape and PD current waveform using K-mean clustering technique. For this purpose, a conventional PD measurement technique given in IEC 60270 standard is adopted to record the PD signals under several insulation faults in the switchgear. Also, a PD current probe is used to detect their corresponding PD current signals. The characteristics of the detected PD signals were studied to estimate the severity of insulation faults. The significant features of time domain PD pulse shape have been compared for the separation of multiple insulation faults in switchgear. Therefore, eight-dimensional feature space has been developed. Finally, K-mean clustering algorithm has been implemented based on the extracted features of PD pulse shape and PD current waveform to define their clusters in the feature space. It has been revealed that the proposed framework is effective in separating the mixed data of PD faults in metal-clad switchgear.