Correlation between two frequency components of various PD pulses in GIS

Detection of partial discharge (PD) signals is one of the most important methods for predictive maintenance of gas insulated switchgears (GIS). Because type of PD source, configuration of a GIS tank and frequency response of a sensor influence a PD spectrum, detected PD pulse has a specified frequency spectrum. Although frequency spectrum is one of valuable information for PD in a GIS, the spectrum measured with spectrum analyzer or FFT processing is generally a result of averaging of many signals so that statistical features of PD are lost. In addition, frequency spectrum is complex and special experience is needed to recognize it. Thus convenient method is required to discriminate various PD pulses. In this paper, we attempt to extract two frequency components from a PD pulse and to process them in order to categorize it easily. Correlation between pulse heights of two frequency components and pulse height distribution resolved with power voltage phase are shown as effective methods to distinguish the type and the location of PD sources.