从故障树到故障贝叶斯网映射的故障诊断方法  

Fault diagnosis method mapping from fault trees to fault bayesian networks

作　　者：杨昌昊[1] 胡小建[2] 竺长安[1] 

机构地区：[1]中国科学技术大学精密机械与仪器系,合肥230026 [2]合肥工业大学计算机网络系统研究所,合肥230009

出　　处：《仪器仪表学报》2009年第7期1481-1486,共6页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(70631003;70801024);安徽省自然科学基金(070416241)资助项目

摘　　要：针对传统的故障树分析方法在复杂系统故障诊断中存在的局限性,论文在定义了故障贝叶斯网及其概率计算的基础上,建立了故障树事件与贝叶斯网结点、故障树逻辑门与贝叶斯网概率分布之间的映射关系,详细研究了将故障树映射为故障贝叶斯网的转变性质,以及利用映射故障贝叶斯网进行故障诊断过程。最后以一种飞机雷达系统故障诊断为例验证了本文所提出的方法。Aiming at the deficiency of traditional fault tree analysis in fault diagnosis of complex systems, mapping relations from events of fault trees to nodes of Bayesian networks and from logic gates of fault trees to probability distributions of Bayesian networks are established based on the definition of fault Bayesian network and its probability computations. The tramsformation properties, in which the fault tree is mapped into fault Bayesian network, are researched in detail, and the processes used fault Bayesian network to make fault diagnosis are described. Finally the proposed method is validated by the example of fault diagnosis in an aero plane radar system.

关 键 词：贝叶斯网 故障诊断 故障树 映射 信念传播 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]