主冷凝器损伤的模糊随机特性及贝叶斯网络分析  

Damage analysis of the main condenser based on fuzzy random Bayesian networks

作　　者：初珠立[1] 杨自春[1] 梁洁[1] 彭茂林[1] 周宗和[1] 

机构地区：[1]海军工程大学船舶与动力学院,湖北武汉430033

出　　处：《哈尔滨工程大学学报》2012年第10期1217-1222,共6页Journal of Harbin Engineering University

基　　金：武器装备预研基金资助项目(9140A27050109JB1112);总装十一五预研基金资助项目(513270301);教育部优秀人才支持计划项目

摘　　要：主冷凝器损伤中故障模式呈现出模糊随机特性以及多态性特征,传统分析方法难以有效解决.应用模糊随机理论对其进行分析,建立了模糊随机贝叶斯网络模型,同时实现了多态故障模式的概率推理.确定了故障事件模糊参数及顶事件的隶属函数,并利用基于Monte Carlo仿真的随机加权方法对隶属函数进行解模糊化.模糊随机贝叶斯网络推理结果与定数截尾试验的误差在1%之内.对各底事件重要度进行分析,并进一步提出BM敏感度概念,以此来分析顶事件对各底事件的概率敏感性。The fault modes of the main condenser always present fuzzy random and polymorphism characteristics, and the traditional analysis methods cannot solve this problem effectively. The fuzzy random Bayesian networks （FRBN） based on fuzzy random theory cuts down the dimensionality for using of conditional independency sufficiently, and polymorphism probability reasoning is realized at the same time. Fuzzy variable of failures and membership funetion of the top event were settled through random weighting of Monte Carlo emulation. The inference re- sult shows that the error percentage is in 1% between FRBN and the truncation experiment. The BM sensitivity was advanced in base of bottom events importance analysis, which shows the sensitiveness of the top event to all the bottom events. At last, importance degree and sensitivity analysis of bottom events which means the sensibility of the top event present some useful suggestions to the operation and maintenance of the main condenser.

关 键 词：主冷凝器 模糊随机理论 贝叶斯网络 损伤分析 

分 类 号：TK229.7[动力工程及工程热物理—动力机械及工程]