基于Bayesian网络的建筑火灾疏散条件安全性评估方法*

为了对建筑火灾疏散条件安全性进行评估，基于Bayesian网络对疏散条件重要构成要素及评估方法逻辑推理过程进行研究探讨。结果表明:评估网络结构、根节点、中间节点及目标节点之间存在因果关联关系；研究得出根节点先验概率与量化节点条件概率表设定方法；Bayesian网络将风险评估与人工智能分析方法相结合，实现对建筑火灾疏散条件的安全性评估，并可用于识别高风险建筑。

Study on assessment method for safety of evacuation conditions in building fire based on Bayesian network

In order to assess the safety of evacuation conditions in the building fire,the important components of evacuation conditions were studied,and the logical reasoning process of the assessment method was discussed.A comprehensive assessment method for the safety of evacuation conditions based on the Bayesian network was constructed,and the causal association relationship between the assessment network structure,root nodes,intermediate nodes and target nodes was illustrated,as well as the setting method for the conditional probability table of the prior probability of root nodes correlated with the quantized nodes.The results showed that the assessment on the safety of evacuation conditions was an important link in the risk assessment of building fire,there were many influencing factors,and there exited the causal association between them.The Bayesian network could combine the risk assessment with the artificial intelligence analysis method,realize the assessment on the safety of evacuation conditions in the building fire,and identify the high risk buildings.