采用特高频法的GIS典型缺陷特性分析

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

气体绝缘组合电器(GIS)由于具有占地少,免维护和对环境影响小等优点,在电力系统中被广泛使用。为了监测GIS的运行状况,常采用特高频(UHF)法对GIS进行缺陷诊断。通常局部放电识别采用的特征参数往往与电压的相位关系密切,但是有时相位信息不便获得,为了克服传统识别方法的限制,设计了5种典型的缺陷模型来模拟GIS运行中出现的缺陷并通过试验,从特高频信号中提取出9个具有物理意义的特征参数来描述5种放电的典型特征,而且根据放电信号的幅值序列q和时间间隔序列Δt处理得到与相位无关的特征量,再用SPSS进行聚类分析。聚类结果证明这些特征量随着缺陷类型的改变而变化明显,可用于缺陷的模式识别。

GIS plays an important role in power systems for its maintenance-free and less influence to environmental and industrial conditions.In order to monitor the on-line condition,ultra high frequency（UHF） method is widely used for defect diagnosis of GIS.Therefore,five types of typical partial discharge（PD） sources were designed and fabricated to imitate the defects in GIS online.To distinguish the types of partial discharges,some meaningful parameters behind the UHF signal were extracted and discussed.In general,the conventional phase-resolved partial discharge（PRPD） and pulse sequence analysis method is adopted to discriminate partial discharge,but sometimes it may be less convenient to employ because of the absence of the phase information.Consequently,nine comprehensive parameters realized by processing the UHF pulse magnitude sequence q and time interval sequence Δt are introduced.The parameters can describe exactly the physical characteristics of different partial discharges.By means of the parameters,the pattern classification of PD without employing the phase information of the applied voltage could be achieved.The clustering results by SPSS indicate that there is evident difference in the characteristic coefficients.