基于Agent的单车道交通流仿真

Agent-based Simulation of Single-Lane Traffic

基于Agent的交通仿真能够细致的刻画交通实体的决策逻辑和微观行为。本文分析了驾驶员-车辆Agent的三层结构,提出了单车道交通流中驾驶员纵向运动决策树,以及操作层的加速度计算公式,建立了基于Agent的空间连续、时间离散的一维交通流仿真模型。采用周期性边界条件仿真,再现了真实交通流的流量-密度关系和速度-密度关系;针对头车四种典型的速度剖面,对由10辆车组成的车队进行了仿真,后车都能实现安全、稳定跟驰。表明该模型既能再现交通流宏观特性,也能再现微观特性。

Agent-based simulation can represent decision logic and microscopic behavior of entities of traffic system in detail. In the paper the structure of driver-vehicle elements with three levels was analyzed, a decision tree of tactical longitudinal (movement) was put forward, the expressions of accelerate and decelerate were derived, and an agent-based continuous (space,) discrete time 1D traffic simulation model with was built. With periodic boundary condition the flow-density and (speed-density) relation of traffic were reproduced. Given four kinds of speed profiles of the leader, a platoon of 10 vehicles was simulated. Each follower can follow leader with safety and stability. The result shows the model can reproduce realistic macro and micro characteristics of traffic flow.