This thesis dealt with an intelligent algorithm for defects identification by the analysis of partial discharge(PD) pulses in gas insulated switchgear(GIS). Needle-plane electrode, plane-needle electrode, floating metal and crack inside spacer were fabricated to simulate the defects in GIS. Each electrode system was placed at the inside of shielding enclosure and a non-inductive resistor was installed between the electrode system and ground wire.
PD pulses were measured by an oscilloscope with the frequency bandwidth of 1GHz and the sampling rate of 5GS/s. And the data aquisition and signal processing were controlled by a LabVIEW program. Also, the digital filter was designed to eliminate a power frequency and external noise.
The parameters of a single PD pulse in relation with the polarity of power frequency were analyzed depending on defects. The physical shapes were compared by using kurtosis, skewness, and time-base parameters such as rising time, falling time, pulse-width, and maximum voltage. By applying the proposed algorithm, the identification rate were 97% in needle-plane electrode, 96% in plane-needle electrode, 91% in floating metal, and 93% in crack inside spacer. From the results, it was verified that the proposed algorithm can identify the type of defects in GIS.