A research on human emergency evacuation based on revised ACO-CA

In a sudden natural disaster, a large number of people may get detained within a hazardous space. In recent years, the number of such incidents has increased in China, which motivates the research of artificial intelligence-based simulations of emergency rescue and evacuation. This paper proposesam ethodological approach that combined with ant colony optimization and Cellular automata integrating simulation and optimizationto study the complexity and randomness characteristics of human behaviors under the emergency evacuation, for solving an optimal emergency evacuation planning problem in an emergency shelter. The path residual pheromone and heuristic factors of the Ant colony algorithm in this integrated model are treated as personal behavior difference and aggregation, which can be treated as the herb behavior factors and the shortest path first factors,reflecting the randomness and interaction in the process of population evacuation. Through using ant colony algorithm to calculate the transition probability of the interaction among the neighboring cells, and updating the pheromone through taboo list based on local optimal path of the ant colony, the cells can finally finish the simulation of safe evacuation under the principle of optimal transition rules. The method could effectively simulate the delayed population and achieve the simulation of the population distribution in the evacuation area when the natural disaster happens. It would also offer a scientific reference for the research of population emergency evacuation.