智能网联环境下基于谱聚类的车辆群体牵制控制策略  

Spectral Clustering-based Pinning Control Strategy for Vehicle Platoons in Connected and Automated Environments

作　　者：王灿 赵妍[1,2,3] 李林恒 曲栩 冉斌 WANG Can;ZHAO Yan;LI Lin-heng;QU Xu;RAN Bin(School of Transportation,Southeast University,Nanjing 211189,Jiangsu,China;Institute on Internet of Mobility,Southeast University and University of Wisconsin-Madison,Southeast University,Nanjing 211189,Jiangsu,China;Jiangsu Province Collaborative Innovation Center of Modern Urban Traffic Technologies,Southeast University,Nanjing 211189,Jiangsu,China)

机构地区：[1]东南大学交通学院,江苏南京211189 [2]东南大学东南大学-威斯康星大学智能网联交通联合研究院,江苏南京211189 [3]东南大学现代城市交通技术江苏高校协同创新中心,江苏南京211189

出　　处：《中国公路学报》2024年第11期235-248,共14页China Journal of Highway and Transport

基　　金：国家自然科学基金项目(52202408);科技创新2030——“新一代人工智能”重大项目(2022ZD0115600);江苏省研究生科研创新计划项目(KYCX23_0308);东南大学博士研究生创新能力提升计划项目(CXJH_SEU 24183)。

摘　　要：智能网联环境下通过对车辆群体中部分关键的网联与自动驾驶车辆实施反馈控制,能够间接影响人工驾驶车辆的运行,从而在效率、安全等方面实现对整体交通流的优化。基于此提出了一种基于谱聚类的牵制控制(Spectral Clustering Based Pinning Control, SC-PC)策略,旨在优化智能网联环境下车辆群体的微观控制方法与效果,提升交通流整体运行水平。首先,面向智能网联环境下的车辆群体提出了网络模型的构建与牵制控制的定义。其次,综合考虑网络拓扑静态信息与车辆动力学动态信息,提出了一种基于谱聚类算法的关键控制节点识别方法,用于确定牵制控制的实施对象。然后,以安全和效率等多目标为导向设计了针对牵制节点车辆的反馈控制方法。最后,基于真实场景下车辆跟驰行驶的TOD数据集开展数值仿真试验,对比分析不同牵制节点识别方法与不同牵制率下的牵制控制效果,并对所提SC-PC策略的有效性进行了验证。结果表明:相较于其他牵制控制策略,所提出的SC-PC策略能够更加准确地识别出车辆群体中的关键控制节点,在交通振荡、同步性、安全性指标方面分别至少提升5.3%、11.7%和16.0%,所提方法能够在提高交通流抗干扰性的同时,实现效率与安全性的同步优化,可服务于智能网联环境下交通流优化控制问题中资源投入与控制效果的权衡。In connected and automated environments,implementing feedback control on key connected and automated vehicles in a vehicle platoon can indirectly influence the operation of human-driven vehicles to thereby optimize the overall traffic flow in terms of efficiency and safety.Hence,this study proposes a spectral clustering-based pinning control(SC-PC)strategy to optimize the microcontrol effects of vehicle platoons in a connected and automated environment and enhance the overall traffic flow performance.First,a network model and definition of pinning control is proposed for vehicle platoons in a connected and automated environment.Second,by comprehensively considering the static network topology information and the dynamic information of vehicle dynamics,a key control node identification method based on the spectral clustering algorithm is proposed to determine the implementation objects of pinning control.Subsequently,a feedback control method targeting pinning node vehicles is designed,guided by multiple objectives,such as safety and efficiency.Finally,numerical simulation experiments are conducted using the TOD dataset of vehicle following behavior in real scenarios.The effects of different pinning node identification methods and pinning rates on the pinning control were compared and analyzed,and the effectiveness of the proposed SC-PC strategy was verified.The results show that,compared with other pinning control strategies,the proposed SC-PC strategy can more accurately identify key control nodes within the vehicle platoon to improve traffic oscillation,synchronization,and safety indicators by at least 5.3%,11.7%,and 16.0%,respectively.Thus,the proposed method can enhance the anti-interference ability of traffic flow while simultaneously optimizing efficiency and safety to serve as a balance between resource input and control effects in traffic flow optimal-control issues in connected and automated environments.

关 键 词：交通工程 混合交通优化 牵制控制 车辆群体 谱聚类 智能网联环境 

分 类 号：U491.4[交通运输工程—交通运输规划与管理]