中国安全科学学报 ›› 2023, Vol. 33 ›› Issue (12): 176-182.doi: 10.16265/j.cnki.issn1003-3033.2023.12.0879

• 公共安全 • 上一篇    下一篇

基于模糊贝叶斯网络的城市商业综合体火灾风险分析

秦荣水1(), 石晨晨1, 陈超2, 兰猛3, 刘小勇4, 肖峻峰1   

  1. 1 安徽建筑大学 土木工程学院,安徽 合肥 230601
    2 西南石油大学 石油与天然气工程学院, 四川 成都 610500
    3 清华大学 工程物理系,北京 100084
    4 清华大学 合肥公共安全 研究院,安徽 合肥 230601
  • 收稿日期:2023-06-03 修回日期:2023-09-13 出版日期:2023-12-28
  • 作者简介:

    秦荣水 (1988—),男,安徽芜湖人,博士,讲师,主要从事消防安全、建筑安全等方面的研究。Email:

    陈超,教授

    刘小勇,正高级工程师

    肖峻峰,教授

  • 基金资助:
    安徽建筑大学博士启动基金资助(2021QDZ04); 安徽省住建厅软科学研究项目(2023-RK044); 安徽省课程思政教学团队(2020kcszjxtd17)

Risk analysis on fire accident of urban commercial complex based on fuzzy Bayesian network

QIN Rongshui1(), SHI Chenchen1, CHEN Chao2, LAN Meng3, LIU Xiaoyong4, XIAO Junfeng1   

  1. 1 College of Civil Engineering, Anhui Jianzhu University, Hefei Anhui 230601, China
    2 College of Petroleum and Natural Gas Engineering, Southwest Petroleum University, Chengdu Sichuan 610500, China
    3 Department of Engineering Physics, Tsinghua University, Beijing 100084, China
    4 Hefei Institute for Public Safety Research, Tsinghua University, Hefei Anhui 230601, China
  • Received:2023-06-03 Revised:2023-09-13 Published:2023-12-28

摘要:

为有效降低城市商业综合体火灾事故风险,首先根据城市商业综合体火灾事故的演化过程及其影响因素的因果分析,构建相应的故障树(FT)和事件树(ET)模型,并将其映射为贝叶斯网络(BN),确定影响因素之间的条件关系;然后采用基于专家判断的模糊理论来确定基本事件的先验概率,构建模糊BN(FBN)模型,以克服风险因素失效概率的不确定性;最后利用FBN建立的城市商业综合体火灾事故风险评估模型进行双向推理和敏感性分析,得出导致不同等级火灾事故的关键影响因素。研究表明:加强防火和防烟分区设计、提高防火分隔设施耐火等级、合理安装防火分隔设施、降低防排烟系统故障率,可有效杜绝高损失风险等级火灾事故的发生。

关键词: 火灾事故, 城市商业综合体, 模糊贝叶斯网络(FBN), 风险分析, 先验概率

Abstract:

To effectively reduce the risk of fire accidents in urban commercial complexes, firstly, based on the causal analysis of the evolution process of fire accidents and influencing factors, the corresponding Fault Tree (FT) and Event Tree (ET) models were constructed. Then they were mapped into Bayesian networks (BN) to determine the conditional relationships between influencing factors. Secondly, fuzzy theory based on expert judgment was used to determine the prior probabilities of basic events. The FBN model was constructed to overcome the uncertainty of failure probabilities of risk factors. Finally, the FBN-established risk assessment model was utilized to perform bidirectional inference and sensitivity analysis for fire accidents of urban commercial complexes, identifying the key influencing factors leading to fire accidents at different severity levels. The study indicates that strengthening fire and smoke zoning design, increasing the fire resistance rating of fire separation facilities, installing fire separation facilities reasonably and reducing the failure rate of smoke exhaust systems can effectively prevent the occurrence of high-loss risk-level fire accidents.

Key words: fire accident, urban commercial complex, fuzzy Bayesian network (FBN), risk analysis, prior probability