基于改进多智能体模型的建筑火灾疏散研究  

Research of building fire evacuation based on improved multi-agent model

作　　者：戴剑勇 高文卉 张澳 李佩东 张雨婷 DAI Jianyong;GAO Wenhui;ZHANG Ao;LI Peidong;ZHANG Yuting(School of Resources,Environment and Safety Engineering,University of South China,Hengyang Hunan 421001,China;Hunan Provincial Key Laboratory of Emergency Safety Technology and Equipment for Nuclear Facilities,Hengyang Hunan 421001,China)

机构地区：[1]南华大学资源环境与安全工程学院,湖南衡阳421001 [2]核设施应急安全技术与装备湖南省重点实验室,湖南衡阳421001

出　　处：《中国安全生产科学技术》2024年第7期101-107,共7页Journal of Safety Science and Technology

基　　金：国家自然科学基金项目(51174116)。

摘　　要：为了在建筑火灾等突发事件下合理指导人群有序疏散、减少人员的伤亡,提出1种改进的多智能体疏散模型,结合元胞自动机对疏散环境进行建模,将占据元胞空间的个体视为1个独立的智能体,设计人群疏散行为策略作为演化规则,将元胞及其属性封装扩展为具有自主性的智能体。多智能体疏散算法实现了个体的差异性,较好地体现了个体情绪感染(恐慌)、协作行为、经验度等对疏散行为的影响。研究结果表明:该模型更加接近现实,能直观地体现不同因素对于人员疏散的影响。研究结果可为公共场所安全设计和制定应急预案提供可靠依据与实践参考价值。In order to reasonably guide the orderly evacuation of crowd under emergencies such as building fire to reduce the casualties,an improved multi-agent evacuation model was proposed.The modeling of evacuation environment was conducted combining with cellular automata,namely the individual occupying the cell space was regarded as an independent agent,then the crowd evacuation behavior strategy was designed as the evolutionary rule,and the cell and its attributes were encapsulated and expanded into an autonomous agent.The multi-agent evacuation algorithm realized the difference of individuals,and better reflected the influence of emotional infection(such as panic)among individuals,cooperative behavior,and experience degree on the evacuation behavior.The results show that the model is closer to reality and can intuitively reflect the influence of different factors on evacuation.The research results can provide a reliable basis and practical reference value for the safety design of public places and the formulation of emergency plans.

关 键 词：多智能体 元胞自动机 人员疏散 情绪感染 建筑火灾 

分 类 号：X928.7[环境科学与工程—安全科学]