A Multi-agent Evacuation Model for High-Rise Building Fire Integrating Environmental Perception and Individual Heterogeneity

High-rise residential buildings have complex structures and long evacuation routes, which further exacerbate the challenges in fire safety. Existing fire evacuation studies rarely pay attention to the influence of smoke diffusion and building layout on the visibility of evacuees and also ignore the modeling of heterogeneous human behavior based on composite factors such as psychological, social, and group dynamics. In this study, we construct a multiagent evacuation model for high-rise building fires that integrates environmental conditions, individual cognition, mobility, and information dissemination. First, we employ agent-based modeling and categorize agents into strong cognitive, weak cognitive, altruistic, and self-oriented types to model their internal drives. Second, we establish a vision loss model based on fire environment perception to depict the vision field of the agents. Furthermore, a mobility evolution model is proposed to represent behavioral decisions and movement speed under the influence of internal drives and vision fields. Finally, an information dissemination model is developed based on the decision-making patterns of the agents in complex scenarios. The evacuation model is ultimately formed through the combination of the mobility evolution model and the information dissemination model. In the experimental design, by simulating the combination ratio of different internal drive agents, different vision conditions, and information dissemination methods, the impact of the proposed model on the fire evacuation performance is analyzed. Results indicate that the internal drive model of the agents significantly impacts evacuation efficiency. The vision loss model and the mobility evolution model are the core elements for agents’ precise environmental perception and accurately feedback. Accurate and immediate coordination between official broadcasts and social dissemination can increase the evacuation efficiency by 13%. Therefore, this model provides a theoretical basis for the design of smoke exhaust in high-rise buildings, the development of intelligent evacuation systems, and the optimization of information dissemination strategies.