基于贝叶斯网络的养老院火灾风险评估方法研究

为预防养老院火灾事故,结合事故树法(FTA)和贝叶斯网络法(BN),建立了一套养老院火灾风险定量评估模型。首先,采用事故树法建立潜在的养老院火灾事故场景;其次,考虑到养老院火灾事故场景中不确定因素的影响,将事故树模型转化为贝叶斯网络模型,并结合有人员伤亡的养老院火灾事故发生发展实际优化模型;最后,以某市养老院为例,结合调研、文献及统计数据确定先验概率及条件概率,并用GENIE 2.0软件实现贝叶斯图形化,分析验证该模型逻辑可行性。结果表明:通过该模型和方法,不仅可以预测养老院火灾事故中各场景发生发展概率,还能对各风险因素敏感度和最大致因链进行分析,提高了风险因素辨识和评价的准确性,可以为养老院火灾风险分析和防控提供参考。

A fire risk assessment method of nursing home based on FTA and BN

This paper combines the fault tree method (FTA) and the Bayesian network method (BN) to establish a set of quantitative assessment models of fire risk in nursing homes. We first use the fault tree method to establish a potential fire accident scene in nursing homes. Then, considering the uncertain factors in the fire accident scene, we convert the accident tree model into a Bayesian network model, and combine the factors that cause casualties to optimize the model. Finally, taking a nursing home in a specific city as an example, we determine the prior and conditional probabilities and achieve Bayesian graphics by GENIE 2.0. The analysis verifies the logical feasibility of the model. The model and method can predict the occurrence and development probability of various scenarios in nursing home fire accidents. Also, the sensitivity of various risk factors and the maximum cause chain can be analyzed, with improved accuracy of identification and evaluation of risk factors.