Multi-agent simulation on day-to-day route choice behavior
日常择路行为的多智能体模拟

The day-to-day route choice is a non-cooperative game process with many participants during a long time period. We use the multi-agent simulation system SeSAm to investigate two different route update rules designed for traffic information release and unrelease respectively. Simulation results on a small grid network show that the rule under information release can make route flow evolve to the logit stochastic user equilibrium faster than that under information unrelease. However, the flow pattern may be unstable if travelers put undue compliance on the released information. A stable flow pattern can also be achieved only relying on travelers' perceived experience and learning ability, but the evolution is rather slow. It is thus suggested that the traffic information should be timely released to public meanwhile utilized properly.