基于贝叶斯网络和理想解动态群决策的故障诊断方法  

Dynamic Group Decision-making Fault Diagnosis Method Based on Bayesian Network and TOPSIS

作　　者：姚成玉[1] 陈东宁[2,3] 冯中魁[1] 吕世君[2,3] 

机构地区：[1]燕山大学河北省工业计算机控制工程重点实验室,秦皇岛066004 [2]燕山大学河北省重型机械流体动力传输与控制重点实验室,秦皇岛066004 [3]先进锻压成形技术与科学教育部重点实验室(燕山大学),秦皇岛066004

出　　处：《中国机械工程》2013年第16期2235-2241,共7页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(50905154);河北省自然科学基金资助项目(E2012203015);河北省教育厅科研项目(ZH2012062);秦皇岛市科技支撑计划资助项目(2012021A078)

摘　　要：为综合利用多属性信息和历次故障搜索结果反馈信息进行故障诊断,提出一种基于贝叶斯网络和理想解动态群决策的故障诊断方法。以砼泵分配阀液压系统为例,利用贝叶斯网络对系统进行分析并求解根节点的后验概率和关键重要度;根据本次诊断成功与否对下次最优搜索决策影响程度的大小,定义出启发函数求解启发式信息价值;考虑后验概率、关键重要度和启发式信息价值等因素,利用基于熵权的理想解法求取搜索方案的群体理想解和逆理想解,得到故障搜索最佳方案;考虑历次故障搜索最佳方案对当前搜索方案的影响,最终求得故障搜索的最佳方案序列。该方法克服了单属性决策和群决策方法的不足,提高了故障诊断的可行性和诊断效率。To utilize multi--attribute information and previous results of fault search,a fault diag- nosis method of dynamic group decision--making based on Bayesian network and TOPSIS was pro- posed. Hydraulic system of a concrete pump distribution valve was given for illustration, the Bayesian network was established to calculate the root nodes＇ posterior probabilities and criticality importance. According to the information that the current diagnosis result was correct or not and the influences on the next optimal search decision--making,a heuristic function was defined to obtain the value of heu- ristic information. The root nodes＇ posterior probabilities, criticality importance and heuristic informa- tion values were considered comprehensively, and the group＇s ideal solution and contrary ideal solution of search schemes were obtained by TOPSIS on the basis of entropy weight,then the best fault search scheme was obtained. The previous fault search results＇ influence on the current search scheme was considered,then the best solution sequence of the fault search was obtained finally. The limitations of single--attribute decision making and group decision--making are overcome, and the feasibility and ef- ficiency of fault diagnosis are improved.

关 键 词：故障诊断 动态群决策 贝叶斯网络 理想解法 砼泵分配阀 

分 类 号：TH137.51[机械工程—机械制造及自动化]