融合D-S理论的锅炉人因贝叶斯网络安全性定量分析

摘要

Uncertain information is a common problem in ship boiler safety analysis,the D-S evidence theory fused information of different experts well and a Bayesian Network was built considering human factors,and got conditional probability of the nodes in form of interzones. Then input the weighted average information to the BN model. Contrast with object-oriented BN, FTA shows that this method can fuse information from different experts better,the data is more accurate for project.%针对船用锅炉人因安全性分析中存在的知识不确定性,采用D-S证据理论对多专家信息进行融合并建立考虑人因的贝叶斯网络,得到节点条件概率的区间表示形式.经加权平均后代入贝叶斯网络计算,与面向对象贝叶斯网络和FTA等方法的对比显示,该方法能够更加有效地融合不同专家信息,也更为符合工程实际.