摘要
自动驾驶与网联技术的加速落地,使混合交通逐渐成为常态,传统同质假设下的动态交通分配已难以解释容量波动、拥堵传播以及关键参数变化所触发的非线性现象。综合百余篇代表性文献,系统梳理异质车辆条件下动态交通分配的建模、均衡、演化与求解研究进展,阐明用户均衡-系统最优连续谱的形成机理、关键特性与调控空间。从微-中-宏三种建模粒度比较异质车辆差异,进而引入“自动化水平×网联水平”的解耦视角,探讨数据驱动的实现路径。归纳对比博弈论、变分不等式与多目标三类均衡理论,以及解析、仿真与学习三条求解路径,整理常用评估场景与指标。总结日内-日际双时间尺度下混合交通均衡的存在、稳定与可达等性质及判据,汇总主流研究得出的阶段临界阈值,并提炼出现有研究的三类局限:以二元划分建模为主,偏重单一效率目标,对演化与调控机制刻画尚浅。未来研究可聚焦“自动化水平×网联水平”二维建模框架、多目标动态均衡与“闭环为主、开环为辅”的协同调控三大方向,以提升动态交通分配的机制解释力与现实适应性,为路网管理、专用车道配置以及出行诱导与定价等政策与工程实践提供科学依据与评估参考。
The accelerated deployment of automated driving and connected vehicle technologies is making mixed traffic the norm.Traditional dynamic traffic assignment(DTA)models,based on homogeneous assumptions,struggle to explain capacity fluctuations,congestion propagation,and nonlinear phenomena triggered by changes in key parameters.Drawing upon more than one hundred representative studies,this paper systematically synthesized recent advances in the modeling,equilibrium,evolution,and solution methods for DTA under heterogeneous vehicle conditions.It elucidated the formation mechanism,key characteristics,and regulatory potential of the user equilibrium-system optimum continuum.Differences among heterogeneous vehicles are compared across micro,meso,and macro-level modeling scales.A decoupled perspective based on the“automation level×connectivity level”framework was introduced,and data-driven implementation pathways were discussed.The study summarized and contrasted three classes of equilibrium theories-game theory,variational inequalities,and multi-objective optimization-along with three corresponding solution approaches:analytical,simulation-based,and learning-driven.Commonly used evaluation scenarios and metrics were also summarized.Furthermore,the paper reviewed the existence,stability,and attainability of mixed-traffic equilibria across within-day and day-to-day time scales,compiling critical phase-transition thresholds identified in the literature.Three major research gaps were a predominance of binary partition modeling,an overemphasis on single efficiency objectives,and an underdeveloped depiction of evolutionary and control mechanisms.Future research should focus on a two-dimensional modeling framework integrating automation and connectivity levels,multi-objective dynamic equilibrium,and coordinated control strategies characterized by“closed-loop dominance and open-loop support”.These directions are crucial for enhancing the mechanistic interpretability and real-world adaptability of DTA theories.The synthesized findings provide a scientific basis and evaluation reference for practical applications such as network management,dedicated lane configuration,and travel demand management through information provision and pricing.
作者
夏新海
邓浩彭
唐俊杰
詹琳
雷虎
易燕
王首硕
XIA Xinhai;DENG Haopeng;TANG Junjie;ZHAN Lin;LEI Hu;YI Yan;WANG Shoushuo(Guangdong-Hong Kong-Macao Greater Bay Area Shipping Research Institute,Guangzhou Maritime University,Guangzhou 510725,China;School of Future Transportation,Guangzhou Maritime University,Guangzhou 510725,China;School of Innovation and Entrepreneurship,Guangzhou Maritime University,Guangzhou 510725,China)
出处
《交通科学与工程》
2025年第6期1-19,共19页
Journal of Transport Science and Engineering
基金
国家自然科学基金项目(52402423)
广东省普通高校重点领域专项(2022ZDZX1021、2024ZDZX1033)
广东省科技创新战略专项资金资助项目(PDJH2024A289)
广州市教育局科研项目(2024312023)。
关键词
智能交通
动态交通分配
混合交通流
交通均衡演化
智能网联车辆
综述
intelligent transportation
dynamic traffic assignment
mixed traffic flow
traffic equilibrium evolution
connected and automated vehicle
review
