基于宏微观耦合模型的城市道路交通流在线估计

实时可靠的交通流估计是城市交通管理与控制的基础.宏观的MCTM模型不 能获取引道路段的微观信息,微观的Paramics 仿真则需路网OD的准确估计, 为避开单一 模型使用的缺陷，本文提出建立宏微观耦合模型.在模型估计的单位间隔内，先利用 MCTM估计基本元胞有效密度和引道元胞初步密度，并在接口处计算仿真发车数量；再 转用Paramics 进行引道微观仿真，利用仿真检测数据计算交叉口排队长度和引道元胞有 效密度，取代初步密度，作为下一个间隔计算的初始输入,实现交通流的在线估计.仿真中, 为符合转向需求实时变化特性，建立基于约束卡尔曼滤波的转向需求估计模型，实时更 新单位间隔的转向需求.实例分析结果表明，宏微观耦合模型满足城市道路交通流在线估 计要求.

Online Prediction of Urban Traffic Flow Based on Macro-micro Model

The reliable and real- time traffic flow prediction is the foundation of traffic management and control in urban traffic. It is difficult for a modified cell transmission model (MCTM) to obtain micro information of the approach section, and also for Paramics simulation model to estimate accurate OD matrix of the whole road network. So a macro-micro model is proposed to keep online traffic flow prediction away from those defects. In unit interval of the prediction, it uses MCTM to predict effective density of basic cells and initial density of approach cells firstly. Then it establishes an interface to calculate simulation vehicle number and uses Paramics for a micro simulation of approach section. The simulation data is used to predict queue length of the intersection and effective density of approach cells to replace the initial ones to be initial input in next interval. During the simulation, a turning traffic demand prediction model based on constrained Kalman filter is established to get the real- time turning traffic demand in unit interval. The simulation analysis indicates the macro-micro model meets the requirements of online traffic flow prediction in urban traffic.