基于贝叶斯网络的多态故障树分析方法  

Multi-state Fault Tree Analysis Method Based on Bayesian Networks

作　　者：周忠宝[1] 马超群[1] 周经伦[2] 董豆豆[2] 

机构地区：[1]湖南大学工商管理学院,湖南长沙410082 [2]国防科技大学信息系统与管理学院,湖南长沙410073

出　　处：《数学的实践与认识》2008年第19期89-95,共7页Mathematics in Practice and Theory

基　　金：十五国防预研项目(41319020103)

摘　　要：针对多态系统故障树分析的难点,通过一个多态雷达系统的实例给出了一种基于贝叶斯网络的多态故障树分析方法.首先根据多态故障树的结构建立贝叶斯网络的拓扑结构,然后根据多态逻辑算子对其进行定量化,进而利用贝叶斯网络分析多态故障树顶事件概率、部件重要度及其它结果.最后通过对实例的分析说明了基于贝叶斯网络的多态故障树分析方法有更强的建模分析能力.Due to the difficulty of multi-state fault tree analysis, a new method based on Bayesian networks is proposed by means of an example of multi-state radar system. The topology of the Bayesian networks is constructed according to the structure of multi-state fault tree, the conditional probability distributions and prior distributions are obtained according to multi-state logic operators. Analysis is performed on the Bayesian networks to obtain the probability of top event, importance measures of components and other results. The powerful ability of the new method while modeling and analyzing multi-state systems is illustrated in the end by means the example.

关 键 词：贝叶斯网络 多态系统 多态故障树 多态逻辑算子 故障诊断 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]