Abstract: In this paper, a real-time dynamic model of multiphase traffic flow is established for an urban junction, and the optimal phase timings and phase orders are developed according to the different traffic flows in different time of a day. Meanwhile, an optimal real-time control
method with genetic algorithm (GA) is presented. The traffic flow in next cycle is estimated via linear pre-estimation method, and the maximum traffic capability is used as the control capability index. so the next cycle timing schedule is determined based on the real-time measured traffic flows in current cycle and the cycle just before. The simulation experiment is carried out with MATLAB language, and the simulation results show that the real-time optimized capability is satisfactory.

Key words: traffic flows, GA, multiphase optimization, dynamic control

摘要： 针对单交叉路口多相位交通流建立一种实时动态模型，并根据不同时刻的不同车流量，
选择最优相位配时和相位顺序.提出一种实时遗传算法优化配时控制方法，通过对本周期及前
一周期的车流量进行实时测量，采用线性预估方法，对下一周期的车流量进行预估，以最大通行能力为路口模型控制性能指标，从而确定下一周期的相位配时方案.采用MATLAB语言编程
进行实时仿真试验，结果表明，算法实时性好，运行稳定.

关键词: 车流量, 遗传算法, 多相位优化, 动态控制