Research on the Typical Partial Discharge Using the UHF Detection Method for GIS

• Published in: IEEE transactions on power delivery 2011-10, Vol.26 (4), p.2621-2629
• Main Authors: Gao, Wensheng, Ding, Dengwei, Liu, Weidong
• Format: Article
• Language: English
• Subjects: Applied sciences; Classification; Connection and protection apparatus; Defects; Discharge; Electrical engineering. Electrical power engineering; Electrical power engineering; Exact sciences and technology; Gas insulation; Gas-insulated switchgear (GIS); Miscellaneous; Monitors; On-line systems; Online; partial discharge (PD); Partial discharges; PD clustering; PD magnitude sequence; PD time interval sequence; Power networks and lines; Studies; Switchgear; Testing. Reliability. Quality control; typical PD sources; UHF; UHF technology; ultra-high frequency (UHF); Voltage
• ISSN: 0885-8977; 1937-4208
• DOI: 10.1109/TPWRD.2011.2166089

Gas-insulated switchgear (GIS) plays an important role in power systems, not only because it is maintenance free and has less of an influence on environmental and industrial conditions, but also due to its low space requirements and high reliability. In order to monitor the online condition of GIS and diagnose existing defects, the ultra-high frequency (UHF) methods are widely used for detecting partial discharge (PD) owing to its sensitivity and anti-interference capability. In this paper, five types of typical PD sources are designed and fabricated to imitate the defects in GIS online. In order to distinguish different types of PDs, some feature parameters from the UHF signals are extracted and discussed. Generally, the conventional phase-resolved PD (PRPD) and pulse-sequence analysis methods are adopted to discriminate PD sources. However, sometimes it is not convenient to employ these methods because of the absence of phase information. In this research, nine feature parameters realized by processing the UHF pulse magnitude sequence q and time interval sequence Δ t are introduced. The parameters can represent exactly the physical characteristics of different PDs. Based on these parameters, a new approach is developed through which pattern classification of PD can be achieved effectively without employing the phase information of the applied voltage. The clustering results obtained by performing statistical product and service solutions indicate that there is evident difference in the characteristic coefficients.