故障诊断的时间因果贝叶斯网模型  

Temporal Causal Bayesian Network Model for Fault Diagnosis

作　　者：周曙[1] 王晓茹[1] 钱清泉[1] 

机构地区：[1]西南交通大学电气工程学院,四川成都610031

出　　处：《铁道学报》2011年第11期70-75,共6页Journal of the China Railway Society

基　　金：国家自然科学基金项目(90610026)

摘　　要：在一类诊断所依据的可观信号(警报)具有时序特性的故障诊断问题中,诊断所依据的警报、这些警报出现的先后顺序以及它们之间的时间间隔,都与诊断结果有关联。针对这一类故障诊断问题,提出了时间因果贝叶斯网模型,采用模糊方式对故障与警报之间的时间因果关系进行离散化处理,用模糊运算来合成多个时间因果关系,通过概率计算获得最大可能的故障假说。理论与算例表明该方法有效可行。The alarms observed,the sequence of these alarms and the intervals between these alarms are critical factors in a certain type of fault diagnosis specially based on alarms of timing characteristics.In this paper a temporal causal Bayesian network model is proposed for this kind of fault diagnosis.The fuzzy method is used to discretize the temporal causal relations between faults and alarms,fuzzy operation is used to combine these temporal causal relations,and the fault hypothesis with the maximum likelihood is obtained by probability calculation.The theory and example demonstrate that this approach is correct and feasible.

关 键 词：故障诊断 时间因果 贝叶斯网 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]