Abstract: :In order to solve escape-route planning for coal mine emergency rescue in China, based on actual data of Shaqu Mine of Huajin Coking Coal Group in Shanxi Province, a simulation model of coal mine accident escape was constructed by using multi-agent simulation platform, which consists of four types of agents, namely exit perception, blind follower, environmental perception and experienced ones. Then, four kinds of escape models, position selection, direction selection, adjacent matrix and breadth first algorithm, were selected according to agent types to generate different escape routes. Finally, two scenarios of chamber construction and exit increase were simulated targeting at escape from fire and temporary collapse accidents of coal mines. The results show that the proposed model can fully simulate escape processes of four kinds of miners' agents, and escape routes generated from breadth first algorithm can help agents quickly find the shortest path. Escape number of miners for four kinds of routes is 58.5% higher on average when chamber is constructed at a perfect place than that with no chamber and 45% higher when exits are increased than that with a single exit.

Key words: coal mine safety, agent-based model, safety escape simulation model, breadth first algorithm, fire escape