基于聚类分析的城市交通TOD优化控制方法

为了降低环形线圈车辆检测器故障率, 基于指数平滑异同移动平均线法对缺失历史数据进行修补, 运用Ward最小方差法对历史交通流量数据进行聚类分析, 以改进立方群准则作为聚类终止条件, 确定TOD多方案控制的最优方案数和最佳切换时刻, 利用交通信号配时优化软件Synchro对TOD优化控制方法进行仿真验证。验证结果表明: 优化控制方法能够提供更精细的TOD控制方案, 更能体现对实际交通需求波动的响应, 优化后控制方案的车均延误减少率平均为11.90%, 其中早高峰前时段的车均延误减少率为20.27%, 晚低峰、晚高峰和早高峰的车均延误减少率分别为12.99%、8.07%、6.25%。

TOD optimal control method of urban traffic based on clustering analysis

To overcome the high failure rate of circular loop vehicle detector in real road, the loss historical data were mended by using moving average convergence divergence method.The historical traffic flow data were clustered by using Ward least square method.A clustering terminal condition was proposed to determine the optimal control plan number and switch time of TOD multi-schedule control based on the modified cubic clustering criterion.TOD optimal control method was simulated and verified by using signal timing optimization software Synchro.Verification result indicates that TOD optimal control method can provide more detailed TOD control plan, which can also respond the fluctuation of real traffic demand.The average decrement rate of each vehicle delay based on the optimal control method is 11.9%, in which the decrement rate in pre-morning peak period is 20.27%, and the values in evening low peak period, evening peak period and morning peak period are 12.99%, 8.07% and 6.25%, respectively.