摘要
气体绝缘金属封闭式组合电器(gas insulated switchgear,GIS)局部放电检测对保证GIS的安全可靠运行具有重要的意义。为了对高压GIS缺陷故障进行有效诊断,试验设计了四种典型缺陷模型,并用超高频法提取局部放电信号,得到Ф-q,Ф-n等分布图谱,获得了能够反映局部放电特征的偏斜度γSk、陡峭度ξKu和局部峰值个数Pe等特征参数。根据所提取的四种典型缺陷信号的特征参数特点,通过模糊K近邻分类(fuzzy K-NN classifier,FK-NN)算法对典型缺陷局部放电信号进行了模式识别。结果表明:当近邻个数K=7、调整参数β=0.75时,FK-NN算法对GIS内缺陷识别能达到较高的识别效果。
Detection for partial discharge of gas insulated metal enclosed switchgear and control gear is of important signifi-cance to ensure safe and reliable operation of GIS.For conducting effective diagnosis on defect fault of high voltage GIS, this paper designs four kinds of typical defect models and uses ultrahigh frequency method to extract partial discharge signals in order to get distribution maps ofΦ-q andΦ-n and acquire characteristic parameters including degree of skewnessγSk , steepnessζKu and partial peak numbers Pe.According to features of characteristic parameters of the four typical defect sig-nals,it conducts mode discrimination on typical defect partial discharge by using fuzzy K-nearest neighbors algorithm.The result shows that when the neighbor number K=7 and adjustment parameterβ=0.75,FK-NN algorithm is able to realize higher effectiveness for discriminating inner defect of GIS.
出处
《广东电力》
2014年第1期81-84,109,共5页
Guangdong Electric Power