基于交通预测的多态交通流信号控制

Multi-state Flow Signal Control Based on Traffic Prediction

为解决现有信号控制方法对多态交通流交叉口适应性不足的问题,通过在交叉口设置多功能进口车道和车辆检测器、车道控制器等硬件设施,进行了短时交通流预测基础上的多态交通流条件下交叉口信号配时优化研究。根据交通量预测数据,建立信号控制延误估计模型,以交叉口总延误最小为优化目标选取多功能车道流向,根据每相位最大排队长度逐步优化绿灯时长并实施信号控制。短时交通预测以小波分析为基础,采用RBF神经网络及Markov链分别预测交通流的稳态与随机部分。使用VISSIM软件对设置多功能车道的交叉口多态流信号控制方法进行了交通仿真。分析结果表明:该方法可有效降低行车延误,提高交叉口服务水平。

The current signal control methods are insufficient in adaptability to multi-state traffic flow intersection. In order to solve this problem, on the basis of short-term traffic flow prediction, signal timing optimization for multi-state flow intersection was studied by setting multi-functional entrance lanes, vehicle detectors, lane controllers and other hardware facilities. According to the prediction data of traffic volume in intersection and signal control delay estimation model, multi-function lane flow direction was selected with the minimum total operation delay as the optimization target. Green light length was optimized gradually based on maximum queue length in every phase and signal control was implemented. Based on wavelet analysis, steady and random part of short-term traffic flow was predicted with radial basis function （RBF） neural network and Markov chain respectively. The multi-state flow signal control method at the intersection with multifunctional lanes was simulated by software VISSIM. The results show that this method can decrease the traffic delay effectively and improve service level of intersection.