基于贝叶斯网络的化工企业人员应急疏散研究  

On the emergency evacuation of the personnel in the chemical enterprises based on the Bayesian network

作　　者：朱敬聪 朱晓光 关磊[1] ZHU Jing-cong;ZHU Xiao-guang;GUAN Lei(Center for Major Hazards Installation Monitoring and Control,Accident Investigation and Identification,China Academy of Safety Science and Technology,Beijing 100012,China)

机构地区：[1]中国安全生产科学研究院重大危险源监控与事故调查分析鉴定中心,北京100012

出　　处：《安全与环境学报》2021年第2期739-745,共7页Journal of Safety and Environment

基　　金：国家重点研发计划项目(2017YFC0804900);中国安全生产科学研究院基本科研业务费专项资金项目(2019JBKY06)。

摘　　要：在化工行业事故中,应急疏散是减少人员伤亡的重要屏障。为分析化工企业火灾事故情景下人员应急疏散成功概率,首先运用鱼骨图对影响人员应急疏散的因素进行定性分析,在此基础上构建贝叶斯网络模型。基于该模型,通过资料数据统计、现场调研和文献查阅得到先验概率,通过三角模糊计算条件概率,进而运用Netic软件定量分析方法计算得到人员应急疏散成功概率及失败后各影响因素发生概率。在此基础上运用动态贝叶斯网络的理念分析人员应急疏散的后验概率。最后,通过敏感性分析得出影响人员应急疏散的薄弱环节,为提高人员应急疏散成功率提供依据。结果表明,人员自身的应急能力对应急疏散影响较大,应急演练是提高应急成功率的主要措施,同时,提高人员应急疏散成功率需要时常更新应急疏散的硬件和软件。The given paper intends to introduce a fishbone diagram to the analysis of the success probability of the emergency personnel evacuation in the fire accident spots of the chemical enterprises.And,for the said purpose,it has done a qualitative analysis of the factors affecting the emergency personnel evacuation in the paper and divided the effects of the involved factors into different categories.In accordance with the above said factors,the Bayesian network model can be constructed.After that,the prior probability can be gained by the data statistics,the literature review and the on-site investigation,so as to obtain the contributive nodes by the triangular fuzzy calculation based on the expert analysis.And,then,the Bayesian conditional probability calculation method can help to gain the condition probability.And furthermore,the probability of the successful and unsuccessful emergency evacuation can be analyzed and calculated through the net quantitative analysis.At the same time,the probability of each influential factor can be diagnosed and worked out.On the above said basis,the probability of the post-mortem evacuation of the personnel can be predicted by calculating the historical data on the basic events with the concept of the dynamic Bayesian network.And,it would be possible to find and work out the weak links affecting the emergency evacuation of the personnel eventually,through the analysis of the sensitivity.Thus,it can be seen from the final calculation results that the residents’own emergency response capacity has a greater impact on the emergency evacuation manner.For example,the escape-taking promptness can be taken as one of the residents’own emergency response capacity,which may have the greatest impact on the personnel emergency evacuation.At the same time,according to the arrangement from large to small,the impact of the remaining intermediate nodes can be taken as a size of the escaping measures taken,personnel safety awareness,evacuation organization,site environment and weather conditions

关 键 词：公共安全 贝叶斯网络 应急疏散 鱼骨图 结果诊断 概率更新 

分 类 号：X937[环境科学与工程—安全科学]