基于贝叶斯网络的故障树定量分析法研究  

On Quantitative Analysis of Fault Tree Based on Bayesian Network

作　　者：张超[1] 马存宝[1] 胡云兰 许家栋[3] 

机构地区：[1]西北工业大学航空学院,西安710072 [2]中船重工集团第705研究所,西安710075 [3]西北工业大学电子信息学院,西安710072

出　　处：《弹箭与制导学报》2005年第S3期235-237,共3页Journal of Projectiles,Rockets,Missiles and Guidance

基　　金：西北工业大学研究生创业种子基金资助(Z20040002)

摘　　要：贝叶斯网络是进行不确定性概率推理的强有力工具。文中实现了故障树向贝叶斯网络的转化,并基于链规则和桶排除法,提出了故障树定量分析的贝叶斯网络新算法,计算得到了顶事件发生的先验概率和基本事件的后验概率,避免了复杂的最小割集和不交化求解过程,显著地减少了布尔代数运算量。算例分析验证了新算法的工程实用性。Bayesian Network (BN) provides a robust probabilistic method of reasoning under uncertainty.The Fault Tree (FT) structure converted into the BN format is achieved,and a new BN algorithm of fault tree quantitative analysis based on the ideology of chain rule and bucket elimination is presented in this paper.The prior probability of the top event and the posterior probability of all basic events are calculated.The complicated minimal cut sets and the disjoint process solving are avoided,and the Boolean algebra computation is reduced obviously.The engineering practicability of the new algorithm is proven by a example.

关 键 词：贝叶斯网络 故障树 链规则 桶排除法 

分 类 号：TB112[理学—数学]