A particle swarm optimization based behavioral and probabilistic fire evacuation model incorporating fire hazards and human behaviors.

摘要

This thesis focuses on simulation of human behaviors under the emergency of building fires using a behavioral model called Vacate . Many types of behaviors are included in this research such as automatically detecting the fire hazards, random walking behaviors in pre-evacuation time, wandering behaviors when occupants are in the environment of heavy smoke, wall-following behaviors in the event of heavy smoke, group evacuation behaviors when evacuation leaders are present, and passive evacuation behaviors, such as turning back to the previous room or other rooms to gather belongings or seek refuge. All of these behaviors are based on the heuristic optimization technique of Particle Swarm Optimization (PSO), which originates from simulating a flock of birds.; This thesis includes the simulation of the decision-making process of the occupants' response to fire. Probabilistic factors are considered while the decisions are mainly based on the combined effects of environment, configuration, procedure, and behaviors.; This thesis also focuses on improving the current fire hazard model by combining the hazard effects of smoke, heat, and toxicity. (Abstract shortened by UMI.)