基于知识约简的故障案例重用方法  

Research on Fault Case Reuse Based on Knowledge Reduction

作　　者：赵皑[1] 柳玉[1] 贲可荣 

机构地区：[1]海军工程大学计算机工程系,湖北武汉430033

出　　处：《武汉大学学报（理学版）》2011年第6期523-529,共7页Journal of Wuhan University:Natural Science Edition

基　　金：国家预研项目(513270104)资助

摘　　要：案例重用是使用旧的经验去解决新的问题,可以提升基于案例推理的故障诊断系统解决问题的能力,本文提出一种基于知识约简的故障案例重用方法,主要包括针对静态案例的值约简算法(improved attribute value reduction algorithm based on discernibility matrix,IAVRADM)以及新增案例情况下的值约简增量式更新算法(incremental updating algorithm for attribute value reduction,IUAAVR),其中,IAVRADM算法利用吸收律和互信息增量作为启发式信息,改进了已有算法中分辨矩阵构造和搜索过程,提高了决策规则的生成效率,IUAAVR算法分析总结出需要进行规则更新的3种情况.实验结果表明,IAVRADM算法的适应性强、时间开销低,IUAAVR算法在获得相同数量的决策规则情况下与案例的数量无关且更具时间优势.Methods of case reuse try to use old experiences to resolve new problems,which can improve capabilities of fault diagnosis system based on case reasoning to answer new problems.In this paper,a reusing method of fault case based on knowledge reduction is given,which includes two algorithms： IAVRADM（improved attribute value reduction algorithm based on discernibility matrix） and IUAAVR（incremental updating algorithm for attribute value reduction）.The former utilizes absorptivity and mutual information as heuristic means to optimize creation of discernibility matrix and searching procedure,which is able to improve production efficiencies of decision rules.The latter gives three conditions to update decision rules.Experiments show that IAVRADM has stronger adaptability,lower time cost,and IUAAVR can reduce time consumption under decision rules with the same quantities as IAVRADM.

关 键 词：基于案例推理 案例重用 故障诊断 知识约简 分辨矩阵 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]