交通信号控制参数的仿真优化方法研究

为优化区域交通网络中各信号控制器的配时方案，利用递推最小二乘算法(RLS)和同时扰动随机近似(SPSA)算法，由检测器流量估计DynaCHINA动态网络交通仿真与分析系统的动态OD矩阵，输入并标定各路段的速度-密度模型参数和饱和流量，获得网络状态的准确估计，包括各路段的速度、密度、流量、队列长度等；在此基础上，利用SPSA算法优化各信号控制器配时参数，包括各信号控制器的周期、相位差和绿信比，使得网络中车辆的平均旅行延误、队列长度、或交叉口通过量等指标最优. 针对实际路网的测试表明，本文的参数标定方法可以获得准确的检测器流量估计，结果明显优于Ashok K的动态OD矩阵与检测器流量估计方法；与现有的基于Synchro信号配时优化软件获得的结果相比较，该方法可较大幅度缩短车辆在路网中的平均旅行延误，并可推广应用于更复杂的区域路网的信号控制参数优化等场合.

Simulation Optimization of Traffic Signal Control Parameters

To optimize the signal controller timing schemes in an regional traffic network, the recursive least square (RLS) algorithm and the simultaneous perturbation stochastic approximation (SPSA) algorithm are developed, which can utilize the surveillance flows to estimate the dynamic OD matrix input and calibrate the speed-density model parameters and saturation flow for each road segment in the DynaCHINA dynamic network traffic simulation and analysis system. By this approach, the network traffic states can be estimated accurately, such as the speed, density, flow, queue length, and so on, for each road segment of the network. Based on the reliable traffic estimation, the SPSA algorithm is proposed to adjust the signal controlling parameters in a network, including the signal cycles, offsets, and splits, so that the network performance index, such as average vehicle travel delay,queue lengths, intersection throughputs, and so on, can be optimized in a dynamic network traffic simulation system. From wide tests for actual network, it is concluded that the proposed method can obtain more accurate estimation of surveillance flows than that of Ashok K’s dynamic OD matrice and sensor flow estimation method, and it can also significantly reduce the average travel delay of vehicles across the network, in comparison with Synchro signal timing optimization software which is now widely used by traffic engineers. Furthermore, the proposed method can be extended to the application of more complicated and large area traffic networks.