摘要
交通状态估计方法能够为交通管理系统提供完备的交通状态信息,为系统管控方案生成、智能车辆自行决策提供数据支持。基于此,面向智能网联环境提出一种基于神经算子的交通状态估计方法。首先,研究人工车辆与具备前瞻能力的网联车辆构成的混合交通流的宏观动力学特性,建立适用于网联环境下的混合交通流模型。其次,考虑交通状态估计的准确性与实时性需求,结合神经算子理论,构建网联环境下交通状态估计模型TSE-DeepONet。最后,基于数值试验与仿真试验对该方法进行测试。研究结果表明:TSE-DeepONet模型在网联环境下能够精准估计研究范围内任意位置、任意时刻的交通状态,估计误差不高于10%;随着渗透率的增加,人工车辆与网联车辆的交通状态估计误差均呈现出下降趋势;网联车辆的前瞻距离及前瞻信息处理策略对其交通状态估计的精度具有显著影响。研究结果可为未来智能网联环境下多尺度协同控制技术的发展提供理论依据和方法支撑。
Traffic state estimation methods are essential for providing comprehensive traffic state information to traffic management systems,and provide data support for the autonomous-generation of control schemes and self-decision-making of intelligent vehicles.Towards intelligent and connected environments,a traffic state estimation method base on neural operators was proposed.Firstly,this paper focused on the macroscopic dynamic characteristics of mixed traffic flow composed of human-driven vehicles and connected vehicles with look-ahead capabilities,and developed a mixed traffic flow model that is suitable for connected environments.Secondly,considering the accuracy and real-time requirements of traffic state estimation,this paper integrated the mixed traffic flow model with neural operator theory to construct a traffic state estimation model,denoted as TSE-DeepONet.Finally,the efficacy of this model was verified through numerical and simulation experiments.The results show that:The TSE-DeepONet is capable of accurately estimating the traffic state at any location and at any time within the study area in a connected environment,with an estimation error no higher than 10%.As the penetration rate of connected vehicles increases,the estimation errors for both human-driven vehicles and connected vehicles exhibit a decreasing trend.Moreover,the look-ahead distance of connected vehicles and their look-ahead information processing strategies significantly influence the accuracy of traffic state estimation.These results provide theoretical foundation and method support for the development of multi-scale collaborative control technology in the future intelligent and connected environments.
作者
陈帅铭
冀淅明
邵海鹏
CHEN Shuai-ming;JI Xi-ming(School of Transportation Engineering,Chang'an University,Xi'an 710064,Shaanxi,China)
出处
《中国公路学报》
北大核心
2026年第1期265-279,共15页
China Journal of Highway and Transport
基金
国家重点研发计划项目(2024YFB4303400)。
关键词
交通工程
交通状态估计
网联环境
神经算子
宏观交通流模型
traffic engineering
traffic state estimation
connected environment
neural operator
macroscopic traffic model
