基于Multi-Agent的多车道交通流的分布式仿真研究

Distributed Simulation of Multi-Lane Traffic Flow Based on Multi-Agent

该文以Agent技术为基础 ,对多车道交通流进行了模拟 ,并与实际的道路交通流特点进行了比较。首先根据驾驶员实际驾车的特点 ,建立车辆Agent的基本模型 ,对车辆Agent换道超车和加速策略进行描述。然后以分布式模型为基础 ,将多车道交通环境中的车辆运行模拟为Multi-Agent的相互作用。在仿真中 ,采用了面向Agent的仿真软件Starlogo。从微观上看 ,所描述的车辆模型能够完成加速和换道过程 ;从宏观上看 ,仿真中交通流特点与实际情况是一致的。说明该文的模型符合实际情况 ,同时也表明了使用Agent技术进行多车道交通流系统仿真的准确性和可行性。

Based on agent technology, the multi-lane traffic flow is simulated and its result is compared with the reality. At first, the vehicle agent is modeled according to the characteristics of the driver, and the lane changing rule and accelerated strategy of the vehicle agent are described. The multi-lane traffic environment is modeled as a distributed system, in which multi-agent simulated as vehicles interact each other. Starlogo, which is oriented to agent, is adopted to perform the simulation. From microscopic side, the described vehicle model can complete the accelerating and lane-changing; form the macroscopic side, the characteristics of the simulated traffic flow is same as the real one. It means that the model introduced in this paper is coincident with the reality, and indicates that it is exact and feasible to simulate the multi-lane traffic environment by using agent technology.