Abstract: With the rapid development of economy,our country will hold more and more mass crowds'activities,e. g.Olympic Games,World Expo.In order to reduce the phenomena of long time queuing and congestion in mass crowds'activities,avoid the population accidents and improve the visitor's satisfacdon,multi-stage dynamic coordination and control model based on multi-agent is proposed. And a coordination algorithm PCI is given. By introducing the concept of Pareto Optimality,a method is given to select the next visiting node and the route choice is made using Logit model.This coordination algorithm is compared with two other existing algorithms by computer simulation. The results show that the coordination algorithm PCI meets the real-time need;the algorithm C and PCI can effectively reduce queuing time;and the algorithm PCI can promote the best social welfare. Moreover,its advantage is more obvious in mass crowds activities.

Key words: mass crowds' activities, coordination, multi-agent, simulation