摘要
电力系统数据的海量与复杂性,决定了其数据具有多层次性,随机性,同时还存在故障信息不完整等特点。针对此问题,以往多采用粗糙集进行约简,并提取相关规则,而当关键信息丢失时,以往的方法并不能导出正确结论,且耗时长。本文提出一种将粗糙集理论与朴素贝叶斯相结合的数据挖掘方法,通过粗糙集求取最小属性约简集,并在此基础上利用朴素贝叶斯诊断出故障概率最大的区,最后针对具体的故障设定值对该方法进行了验证。算例结果表明,该方法能在故障信息不完整甚至丢失核属性的情况下得到较好的诊断结果,提高了系统的容错性。
The data of the electric power system is multi-level, random, and also may be incomplete. This paper presents a data mining method which combines the rough set theory with the Naive Bayesian classification. The minimum attribute reduction set is first extracted by using the rough set method,then the biggest area where the possible fault may come up is diagnosed with the naive Bayesian classification method. An application example is also presented.
出处
《自动化技术与应用》
2009年第3期15-17,共3页
Techniques of Automation and Applications
关键词
电力系统
粗糙集
朴素贝叶斯
属性约简
故障概率
容错性
electric power system
rough set theory
Naive Bayesian
attribute reduction
fault probability
fault tolerance capability
