为更加真实地反映车道变换行为,建立具有良好性能的车道变换模型,根据抽样调查结果分析了车道变换的基本条件。引入智能主体(Agent)理论,将Agent与车道变换行为联系起来,建立了基于多智能主体系统(Multi-Agent System,MAS)的车道变换模型框架;给出了车辆MAS的结构,抽象了各Agent成员的元组构成,给出了各Agent成员的工作原理与执行流程;基于多智能系统工程方法(Multi-Agent Systems Engineering Methodology,MASEM),给出了车道变换的仿真流程并得到仿真实例。研究结果表明:在2 km长的双向4车道高速公路上,在30 s时间内,利用提出的车道变换模型仿真得到的车道变换车辆次数与实际情况相差8%;在该时段内任意瞬间,正在执行车道变换的车辆数目与实际情况相差仅1.67%。

In this paper, the fundamental conditions of lane-changing model are analyzed to make full use of random sampling in order to describe the lane-changing behavior realistically and set up a good performance lane-changing model. Relating agent theory with lane-changing behavior, the frame of lane-changing model based on multi-agent system is constructed. Furthermore, the construction, elements and operation program of agent that compositions of the vehicle multi-agent system are provided. The simulation program of lane-changing is represented with taking advantage of multi-agent systems engineering methodology and an example is made correspondingly. The results of simulation show that the mean error of changed vehicle number compares simulation with reality is 8% and the mean error of changing vehicle number is 1.67% under the condition of four lane freeway and in half minute.