基于贝叶斯网的钢筋混凝土耐久性损伤诊断专家系统研究.pdfVIP

　2005 年 9 月 系统工程理论与实践 第 9 期 　 ( ) 文章编号 2005 基于贝叶斯网的钢筋混凝土耐久性损伤诊断专家系统研究 易伟建 ,李 　浩 (湖南大学土木工程学院 ,湖南 长沙 410082) 摘要 : 　在简要分析钢筋混凝土的耐久性问题及影响因素的基础上 ,通过引入基于贝叶斯网的知识表达 和相应的不确定性推理原理 ,构造了钢筋混凝土耐久性损伤诊断专家系统. 与一般基于规则的专家系统 相比 ,贝叶斯网专家系统利用先验概率分布 ,可以使推理在输入数据不完备的基础上进行 ; 以网络的拓 扑结构表达定性知识 , 以网络节点的概率分布表达知识的不确定性 ,从而使不确定性知识的表达直观 、 明确 ;利用贝叶斯法则的基本原理 ,可以实现由因到果及由果到因的双向推理. 关键词 : 　不确定推理 ;贝叶斯网;耐久性 ;损伤诊断 ;专家系统 中图分类号 : 　TP391. 5 ; TU528. 07 　　　　　　　　文献标识码 : 　A 　　　 Bayesian Network Based Expert System for Damage Diagnosis of the Durability of the Reinforced Concrete YI Weij ian , L I Hao (Civil Engineering Department , Hunan University , Chiangsha 410082 , China) Abstract : 　Based on the concise analysis of the durability and its influencing factors of the reinforced concrete , Bayesian network and corresponding uncertainty reasoning principle has been introduced to develop an expert system for damage diagnosis of the durability of the reinforced concrete . Compared with the regular rule - based expert system , the Bayesian network based ES can reason on the incomplete input information using the prior probability distribution ; the topological structure of the network being used to express the qualitative knowledge and the probability distributions of the nodes in the network being used to express the uncertainty of the knowledge , which made the knowledge representation more intuitively and more clearly ; applying the principle of the Bayesian chaining rule ,