基于DBN的气体泄漏事故情景推演与节点分析  

Scenario deduction and node analysis of gas leakage accidents based on DBN

作　　者：岳文静 杜丽敬[1,2] 陈先锋[1] 袁必和 杨家祺 王朋程 YUE Wenjing;DU Lijing;CHEN Xianfeng;YUAN Bihe;YANG Jiaqi;WANG Pengcheng(School of Safety Science and Emergency Management,Wuhan University of Technology,Wuhan Hubei 430070,China;China Research Center for Emergency Management,Wuhan University of Technology,Wuhan Hubei 430070,China)

机构地区：[1]武汉理工大学安全科学与应急管理学院,湖北武汉430070 [2]武汉理工大学中国应急管理研究中心,湖北武汉430070

出　　处：《中国安全科学学报》2022年第S01期165-170,共6页China Safety Science Journal

基　　金：湖北省重点研发计划项目(2021BCA216)。

摘　　要：化工园区危险化学品气体泄漏事故在演化过程中具有路径复杂和发展动态性高的特点,为解决气体泄漏事故难以快速有效应对的问题,以张家口“11·28”重大爆燃事故为例构建推演模型。首先基于动态贝叶斯方法构建基于情景状态、外部环境和应急措施的灾害链情景演化模型,利用联合概率公式计算情景状态概率,实现气体泄漏事故的动态推演;然后结合复杂网络拓扑结构相关知识,利用综合重要度公式,探究事故情景推演过程中的关键节点;最后提出基于数据参考的针对性断链措施。研究表明:基于动态贝叶斯的情景推演模型可以在事故演化路径中起到预测作用,推演结果为相关事故灾前预测、灾中应急和灾后恢复提供更为科学有效的依据。In order to quickly and effectively respond to gas leakage accidents of hazardous chemicals in chemical parks considering their characteristics of complex paths and high development dynamics in the process of evolution,a deduction model was constructed based on″11·28″major deflagration accident in Zhangjiakou.Firstly,according to dynamic Bayesian method,a scenario evolution model of disaster chain based on scenario state,external environment and emergency measures was constructed,and scenario state probability was calculated by using joint probability formula to achieve dynamic deduction of the accidents.Then,key nodes in scenario deduction were explored based on complex network topology and using comprehensive importance formula.Finally,targeted chain breaking measures were proposed based on data reference.The results show that the scenario deduction model based on dynamic Bayesian can play a predictive role in the evolution path of accidents,and deduction results provide a more scientific and effective basis for pre-disaster prediction,emergency response in disaster and post-disaster recovery.

关 键 词：动态贝叶斯网络(DBN) 气体泄漏 情景推演 关键节点 灾害链 重要度 

分 类 号：X937[环境科学与工程—安全科学]