马成元

入学时间：2017级

答辩时间：2022年

论文题目：城市道路交叉口网联混合交通优化控制方法

中文摘要

摘要

随着车路协同和自动驾驶技术的发展，道路交通网联化为交通控制提供了实时车辆轨迹信息，自动驾驶车辆也使得交通流运行机理发生结构化改变，城市交通控制迎来新的机遇和挑战。而目前关于网联混合交通环境下的交通控制研究尚处于探索阶段，未形成完善的理论和方法。本文即在由网联人类驾驶车辆（Connected and Human-driving Vehicle，CHV）和网联自动驾驶车辆（Connected and Automated Vehicle，CAV）组成的网联混合交通环境下，研究城市道路交通控制在单交叉口和多交叉口层面的基本理论和方法。

首先，解析了城市道路网联混合交通运行机理。研究了网联混合交通环境下CAV在交叉口处轨迹规划及控制策略。考虑车辆横向变道与纵向驾驶耦合关系，基于实时交通状态信息和交叉口信号配时，以车辆自身效率和驾驶平顺性为目标，建立双层规划模型优化CAV横纵向时空轨迹。针对轨迹规划双层模型特点，设计并行蒙特卡洛树搜索算法进行快速求解，并通过滚动规划时间窗实现车辆轨迹动态优化。通过实验算例证明，模型不仅能够提高CAV的运行效率，同时可以使得CAV带领混合交通流平顺高效通过交叉口。

其次，在单交叉口层面，建立了网联自组织混合交通环境下的信号配时优化方法。在不限制固定的交叉口相位组合、相序的灵活信号配时框架下，以车辆总延误最小为优化目标，建立混合整数线性规划（Mixed Integer Linear Program，MILP）模型优化交叉口信号配时。考虑自组织网联自动驾驶车辆（Self-organizing Connected and Automated Vehicle，SOCAV）的自主控制权，基于车辆跟驰模型和SOCAV轨迹规划自私目标假设，预测SOCAV和CHV通过停车线时的状态，进而精准设计信号配时以促进SOCAV带领车队高效通过交叉口的“带头效应”，提高绿灯时间利用效率。针对信号配时优化模型解空间特点，设计并行粒子群搜索算法。提出了一种滚动时间窗优化框架，实现根据交通状况的变化动态更新最优信号配时。通过仿真实验证明，模型在平均车辆延误和交叉口通行能力方面显著优于固定信号配时和感应信号配时控制策略。

再次，在CAV轨迹控制与信号配时优化研究基础上，提出一种基于逻辑的信号和轨迹协同控制方法。根据快慢变量伺服原理，设计了一种交叉口车辆轨迹控制快变量与信号控制慢变量之间协同控制框架，并分别提出基于逻辑的交叉口信号配时与CAV速度控制方法，避免了基于运筹优化的协同控制模型求解复杂度高的问题。采用头车轨迹控制的方法防止混合交通流中人类驾驶车辆驾驶不确定性对控制鲁棒性的影响。通过仿真实验验证了协同控制方法的优势，并在敏感性分析中讨论了混合交通群体控制效率与公平性之间的平衡。

最后，在多交叉口层面，利用个体车辆的实时状态和未来路径信息，分别从“基于规则”和“基于优化”两个角度构建了多交叉口信号协同优化控制方法。基于规则的方法利用竞争-合作的群体决策机制，构建了一种网联环境下多交叉口协同信号配时设计方法。模型兼顾多交叉口协同效益和单交叉口控制优化多目标，将单点信号优化构建为各相位的交叉口通行权的竞争过程，将多点协同构建为上下游相位之间的协作过程。并建立了分层动态决策框架，在单层决策中剥离了上下游交叉口控制决策对本地决策的影响，避免了因为状态和决策的耦合性导致模型复杂度提升。通过交通仿真平台验证了协同控制方法在交通控制效率和稳定性上的优势。基于优化的多交叉口信号配时方法利用网联交通环境下各车辆实时状态和路径信息，以最小化车辆延误和停车次数为目标建立MILP模型，在不限制相序和相位组合的灵活框架下优化多交叉口信号配时。模型中主动预测每辆车在其路径上各交叉口的通过时间，考虑信号配时和其他车辆的影响，以实现上下游交叉口信号配时精准协同。仿真实验证明所提出的信号协调控制方法可以有效地降低车辆在交叉口停车率，显著提高交通运行效率。

本文从交通优化控制问题理论基础出发，解析了网联混合交通环境运行机理。构建了自动驾驶横纵向轨迹规划方法，提出了单交叉口和多交叉口的交通控制方法，初步形成了网联混合自组织交通环境下交通控制基本架构，对后续相关研究有重要的借鉴意义。

关键词：交通控制，车路协同，自动驾驶，网联混合交通，优化控制

英文摘要

ABSTRACT

With the development of vehicle road cooperation and automated vehicle technologies, connected vehicles provide real-time vehicle trajectory information for traffic control, and automated vehicles also make structural changes in the traffic flow dynamics. Urban traffic control faces new opportunities and challenges. The research on traffic control under mixed connected traffic environment is still a cutting-edge research topic, and related theory and method are still developing. This paper studies the basic theories and methods of urban road traffic control at isolated intersections and multi-intersections under the mixed connected traffic environment composed of Connected and Human driving Vehicle (CHV) and Connected and Automated Vehicle (CAV).

Firstly, the dynamics of mixed urban traffic are analyzed. The trajectory planning strategy of CAV at intersections under mixed traffic environment is studied. Considering the coupling relationship between vehicle lateral lane changing strategy and longitudinal driving strategy, a bi-level optimization model is established to optimize the CAV trajectory with the objective of efficiency and driving smoothness based on real-time traffic status information and intersection signal timing. According to the characteristics of the bi-level trajectory planning model, a parallel Monte-Carlo tree search algorithm is designed to solve the model efficiently, and a rolling horizon scheme is designed to realize the dynamic optimization of vehicle trajectory. The numerical results show that the proposed model can not only improve the operation efficiency of individual CAV, but also enable CAVs to lead the mixed traffic flow to pass the signalized intersections smoothly and efficiently.

Secondly, at the isolated intersection level, a signal timing optimization method under the mixed connected traffic environment is established. Under the “structure-free” signal timing framework without fixed phase combination and phase sequence, a Mixed Integer Linear Program (MILP) model is formulated to optimize the signal timing to minimize total vehicle delay. Considering the autonomous control of Self organizing Connected and Automated Vehicle (SOCAV), based on the car following model and the assumption of self-interested trajectory planning of SOCAVs, the states of SOCAVs and CHVs passing the stop bar are predicted. The signal timings are accurately designed to utilize the “leading effects” of SOCAVs that SOCAVs can lead the mixed platoons to pass the intersection in an effective way. According to the characteristics of the solution space of the proposed signal timing optimization model, a parallel particle swarm optimization algorithm is designed. A rolling horizon framework is proposed to dynamically update the optimal signal timing according to the varying traffic conditions. The simulation results show that the model significantly overperforms the fixed signal control and vehicle-actuated signal control strategies in terms of average vehicle delay and intersection capacity.

Thirdly, a logic-based signal and trajectory coordinated control method is proposed based on the research of CAV trajectory control and signal timing optimization. According to the servo principle of the fast- and slow-variables, a cooperative control framework between the the slow variables, i.e., the signal timing, and the fast variables, i.e., the vehicle trajectory strategy at intersections is designed, and the intersection signal timing and CAV speed control methods are proposed in a logic-based way, respectively. The proposed method avoids high complexity in the model solution of the optimization-based methods. The trajectory control method is just applied to the leading CAV to improve control robustness under the mixed traffic environment considering the driving uncertainty of human driving vehicles. Simulation experiments show the advantages of the proposed cooperative control method, and the balance between the efficiency and equity of mixed traffic is discussed in the sensitivity analysis.

Finally, at the multi-intersection level, using information of the real-time states and future path information of individual vehicles, the coordinated signal control methods are constructed in the "rule-based" way and "optimization based" way, respectively. The rule-based method uses the group decision-making mechanism with competition and cooperation to construct the coordinated signal timing method at multi-intersections under the connected traffic environment. The model considers both the synergy of multi-intersections and the optimal control of isolated intersections. The signal optimization at isolated intersections is constructed as the right-of-way competition process of all the phases, and the coordination among multi-intersections was modeled as the cooperation process between the upstream and downstream phases. A hierarchical dynamic decision-making framework is established to prevent the influence of upstream and downstream intersection control decisions on local decisions, avoiding the increase of model complexity due to the coupling of the state variables and the decision variables. The advantages of the proposed coordinated control method in traffic control efficiency and stability are verified through the traffic simulation platform. In the optimization-based signal timing method at multi-intersections, a MILP model is formulated to optimize the signal timing in a structure-free framework with the objective of the vehicle delay and parking times minimization using the real-time state and future path information of each vehicle under connected traffic environment. The proposed model predicts the passing time of each vehicle at each intersection on its path, and considers the influence of signal timing and other vehicles to achieve accurate coordination of signal timing at upstream and downstream intersections. The simulation results show that the proposed signal coordination control method can effectively reduce the parking rate of vehicles at intersections and significantly improve the traffic operation efficiency.

In this paper, the urban traffic dynamics under connected mixed traffic environment are analyzed from the theoretical basis of traffic optimization control problem. The longitudinal and vertical trajectory planning method of CAVs at intersections under mixed traffic environment is constructed, and the traffic control methods of isolated intersections and multi-intersections are proposed. The basic structure of the urban traffic control under the self-organized connected traffic environment is preliminarily formed, which is of great significance for the subsequent related research.

Key Words:traffic control, vehicle-road cooperation, automated vehicle, mixed connected traffic, optimal control