面向交通控制的实时在线仿真参数标定

提出了在信号交叉口建立面向交通控制的实时在线微观仿真及仿真模型标定的方法.采用车队离散图示作为拟合指标,通过不断调整待标定参数使仿真与实地中车队离散的差异值减少至合理的误差范围.利用粒子群算法,以上海市嘉定区的曹安路嘉松北路交叉口为研究对象进行实地验证,结果表明,该方法仅通过11次迭代便求解出误差在2%以内的参数解,证明该方法可行及有效并为信号交叉口交通控制模型的研发或优劣的检验提供了基础条件.

Parameter Calibration in Real time Traffic Control Microsimulation

A methodology for setting up a real-time traffic control microsimulation and calibrating the microsimulation model at signalized intersection is proposed. The optimum values for calibration parameters were obtained by minimizing the error between the simulated and field platoon dispersion graphs. The methodology has been applied to the intersection of Cao-an Road and North Jia-song Road in Jiading, Shanghai using Particle Swarm Optimization (PSO). Experimental results show that this methodology can work out a valid solution by only 11 iterations which can make its error less than 2%. The methodology proves to be effective and efficient and also provides a basis for testing or developing traffic control models at signalized intersection.