城市道路交通流离散模型优化分析  

Analysis on Urban Traffic Flow Dispersion Based on Vehicle Plate Recognition Data

作　　者：李瑞敏[1] 唐瑾[1] 

机构地区：[1]清华大学交通研究所,北京100084

出　　处：《公路工程》2015年第4期19-23,27,共6页Highway Engineering

基　　金：国家科技支撑计划课题(2014BAG03B03)

摘　　要：研究信号控制干道不同交通流状态下路口间车流离散模型。针对我国特有的机非混行、机动车交通流车辆构成复杂等交通特性,首先基于实际调查数据分析交通流离散特性,然后对Robertson几何分布交通离散模型进行修正,考虑下游到达车流与上游驶出车流的相关关系,提出3种线性回归模型。最后将多个模型的预测效果进行对比分析,结果表明,在所试验的路段条件下,考虑上游对应时段车流到达、上游对应时段的前一时段车流到达、上游对应时段的后一时段车流到达、下游前一时段车流到达的四元线性回归模型在不同的交通流状态下,能够更好地拟合车流离散规律。This paper focuses on prediction of traffic flow dispersion under different traffic state. Considering the special traffic characteristics in China,such as traffic disorder,mixed traffic flow etc. Firstly,this paper analyses the characteristics of traffic flow dispersion based on actual research data. Secondly it corrects the dispersion parameters of geometric distribution model. Thirdly it proves the linear regression model considering the downstream arriving flow and upstream departing flow. Finally,it compares the corrected geometric distribution model and linear regression model,the results show that the linear regression model of four variables fits better with the dispersion.

关 键 词：车流离散 卡口数据 几何分布 线性回归 平均绝对误差 

分 类 号：U491.14[交通运输工程—交通运输规划与管理]