中心分布归类法对短电缆局部放电的模式识别

Automated Recognition System of Partial Discharge Source in Cable Using Centrum Score

研制了一套微机辅助的可对局部放电源进行检测和识别的系统，建立了一种以马氏距离对局部放电指纹特征空间划分为基础的中心分布归类识别法（ＣｅｎｔｒｕｍＳｃｏｒｅ），并编制了识别不同类型放电缺陷的通用计算机程序．该方法对４种典型的交联聚乙烯（ＸＬＰＥ）电缆局部放电缺陷的测试和分析表明：文中的方法和计算程序是可靠的，可用于识别不同的电缆局部放电模型．文中对特征值进行了仔细筛选，从而提高了计算的效率和识别的准确率．

With the aid of computer, partial discharge (PD) sources could be recognized by using multi dimentional discharge patterns. Such information is useful for determining the use of insulation. The centrum score method for classifying PD patterns is made possible by measuring the distance between the unknown discharge and mean value of those already known. Statistical operators were used in the calculation. The centrum score method can efficiently and accurately classify the fingerprints of different PD patterns in crosslinked polyethylene(XLPE) cable.