Abstract: Aiming at the movement behaviors of subway passengers,based on the Agent model and the particle swarm algorithm,through utilizing the simulation platform of Repast Simphony,this paper built a simulation model of crowd movement behavior.The model simulates the process of passengers entering the waiting room of subway station,then looking for subway’s door for queuing,and entering the subway when subway arrives at station.On this basis,this paper proposed an improved routing algorithm based on Markov decision model.Experimental results show that this algorithm can effectively solve the problem that traditional particle swarm algorithm is easy to trap in local solution,and significantly reduce the number of collisions.In addition,this paper proposed a method to avoid congestion in some compartments through employing passenger number indicator of subway compartment.Experimental results show that this method is effective and can improve the efficiency of passengers entering the compartment by 9%.

Key words: Markov, Modeling, Particle swarm, Queuing, Repast dimphony, Simulation