融合多源数据与元胞传输模型的高速公路交通状态估计方法  

Freeway traffic state estimation based on multi-source data and cell transmission model

作　　者：易术 黄丹阳 YI Shu;HUANG Dan-yang(Sichuan Intelligent Transport System Management Co.,Ltd.,Chengdu 610200,China)

机构地区：[1]四川智能交通系统管理有限责任公司,成都610200

出　　处：《交通运输工程与信息学报》2023年第4期103-114,共12页Journal of Transportation Engineering and Information

基　　金：国家重点研发计划项目(2021YFB1600100)。

摘　　要：针对高速公路管控和决策应对交通状态进行准确、可靠和精细化估计的需求,本文提出了一种基于多源数据+元胞传输模型(Multi-Source Data Cell Transmission Model,MD-CTM)的交通状态估计方法。该方法针对传统CTM模型要求元胞长度必须一致的局限性,提出了一种元胞长度划分的优化方法,能够灵活调整元胞长度和数量。同时,应用卡尔曼滤波技术,将ETC门架流量、稀疏视频检测器流量和样本车辆平均速度数据融合,并与CTM模型相结合,实现高速公路元胞级交通状态估计。为了验证本文提出方法的有效性和准确性,我们利用VISSIM软件构建了长度5 km的高速公路仿真场景。仿真案例结果表明,本文提出的MD-CTM模型能够较为准确地反映不同流量需求下交通流状态的时空演化特征,且相较于CTM模型,其元胞密度估计精度提高12.59%~36.26%。此外,本文选取了成都市绕城高速路段实际场景,对模型的运行效果进行了展示。Accurate,reliable,and efficient traffic state estimation is essential for effective freeway management and decision-making.This study presents a traffic state estimation method called MD-CTM,which combines multi-source data and the cell transmission model(CTM).As the traditional CTM has limitations owing to fixed cell lengths,we propose a cell division approach that allows for flexible lengths and numbers.To enhance the accuracy of traffic state estimation,we utilize the Kal-man filtering technique to fuse different types of traffic data,including traffic flow from the electron-ic toll collection system and sparse video detectors,and an average link speed with the CTM to achieve cell-level traffic state estimation on freeways.To evaluate the performance of the proposed approach,we conducted simulations using VISSIM on a freeway section of 5 km.The simulation re-sults show that the proposed MD-CTM model improves the accuracy of cell density estimation by 12.59%~36.26%compared with the CTM model.Furthermore,our model effectively captures the spatio-temporal evolution characteristics of traffic flow states under different traffic demand condi-tions.Moreover,a real-world scenario of Chengdu city is used to further demonstrate the effective-ness of our proposed approach.

关 键 词：智能交通 交通状态估计 卡尔曼滤波 元胞传输模型 多源数据融合 

分 类 号：U495[交通运输工程—交通运输规划与管理]