摘要
本文将关联规则挖掘与模糊推理方法应用到XLPE电缆的局部放电模式识别中,采用竞争聚类方法划分区间以离散化特征,通过关联规则法挖掘特征间的相互关系来提取分类规则,进而将这些规则模糊化用于模式识别。该方法能有效挖掘出各特征参数与缺陷类型的潜在规则,对局部放电的模式识别和电缆绝缘故障诊断具有极大的参考价值。本文针对几种典型的XLPE电缆局放数据,提取相关的统计特征参数,采用该模式识别系统进行分类,并与多层感知神经网络、决策树C4.5等方法识别的结果进行对比分析。实验结果表明该算法提出的规则具有识别率高、识别速度快、解释性好和区间可动态划分等特点,提供了一种局部放电模式识别新的可行方案。
The article applies association rule mining and fuzzy inference mechanism to pattern recognition of partial discharge （PD） in XLPE cable, uses competitive agglomeration method in section division to show its discretization quality, and finds the relation between sections to collect classifying rules by association method. Then, the paper uses the fuzzy rules into pattern recoganition. This method can effectively discover the potential rules among characteristic parameters and the defect type. So it has great reference values for pattern recognition and fault diagnosis. Aiming at several typical XLPE cable discharge data, this paper extractes the relevant statistical characteristic parameters, classifies the defects by this method and makes a contrast analysis with the results from neutral network and C4.5 methods. Experimental results demonstrate the verification algorithm effective, furthermore, this technique has the advantage of better interpretation, time saving and dynamic interval, which makes it a new solution available for the pattern recognition of PD.
出处
《电工技术学报》
EI
CSCD
北大核心
2012年第5期92-98,共7页
Transactions of China Electrotechnical Society
关键词
局部放电XLPE电缆
关联规则
模糊推理
模式识别
Partial discharge, XLPE cable, association rule, fuzzy inference, pattern recognition
作者简介
姚林朋男，1981年生，博士研究生，主要从事大型电力设备的在线监测、智能模式识别及故障诊断的研究。x
徐颖敏女，1986年生，硕士研究生，主要从事大型电力设备的局部放电检测、智能模式识别及故障诊断的研究。
