智能网联车辆全面普及之前，交通系统将在很长一段时间内呈现智能网联车辆与人工驾驶车辆混合行驶的状态（即不完全车联网环境）。由于车辆驾驶行为的不同，不完全车辆网环境下的交通流行为将更加复杂，交叉口依然是路网中的瓶颈。不完全车联网环境下，交通信号控制依然是保证交叉口交通秩序及交通安全的主要手段。交叉口交通信号控制策略的效果取决于交通流模型对交叉口运营性能快速、准确的评估。本文同时考虑交通需求的动态随机性、服务能力的状态相关性以及交通拥挤传播，构建交叉口多类顾客反馈排队网络模型，用以准确描述交叉口的运营性能。并进一步研究了不完全车联网环境下交叉口群交通信号控制理论与方法。本文的主要工作体现在：（1）构建了考虑拥挤传播的交叉口反馈排队网络模型。为研究交通系统中固有的随机性，本文引入排队理论。为将拥挤导致的系统服务能力波动纳入建模，本文引用状态相关排队理论。为描述交通需求的时变性，本文引入了流体排队理论。为描述交通拥堵对交通需求的影响，本文引入了反馈排队机制。通过将交叉口的各类设施抽象为节点排队系统，将信号控制策略描述为交叉口网络的动态拓扑，从而构建了考虑拥挤传播的交叉口反馈排队网络模型，用以准确估计交叉口运营条件。仿真验证结果表明，本文提出的模型在不同交通强度条件下的平均相对误差为6.4337%。（2）基于上述交叉口反馈排队网络，进一步构建了不完全车联网环境下的交叉口多类顾客反馈排队网络模型。考虑智能网联车辆及人工驾驶车辆的跟驰特性，构建不完全车联网环境下的交通流基本图模型。进而，研究不同渗透率条件下混合交通流平均速度与交通密度之间的关系。通过指数速度模型拟合混合交通流平均速度与交通密度之间的关系，用以标定排队系统的服务率。接着，将不完全车联网环境中的不同车辆类型视为不同顾客，基于考虑阻塞传播的交叉口反馈排队网络模型，进一步构建交叉口多类顾客反馈的排队网络模型。仿真验证结果表明，本文提出的模型在不同交通强度条件下的平均相对误差为6.4463%。（3）构建了考虑拥挤传播的交叉口信号配时优化方法。首先，分别推导了交叉口运营性能指标如车辆平均延误时间、平均油耗以及平均排放等。接着，构建了以车辆平均延误时间最小为目标的优化模型。为适应交通需求动态性，设计了滚动优化策略。算例分析的结果表明：1）本文提出的滚动优化策略能有效降低单点交叉口的车辆平均延误。2）随着智能网联车渗透率的增加，车辆最优平均延误时间进一步降低。（4）构建了考虑动态交通分配的区域交叉口信号配时优化方法。考虑到交通信号控制方案与交通需求的交互影响，本文在交叉口群信号优化算法中嵌入了混合交通流动态交通分配算法。其中，智能网联车辆及人工驾驶车辆的路径选择行为分别描述为动态用户均衡及随机动态用户均衡。结果表明，考虑动态交通分配的交叉口群信号配时优化算法能够进一步降低车辆的平均逗留时间，同时能够缓解区域内各排队节点的饱和度。

Before the full popularization of Intelligent Connected Vehicles (ICVs), the transportation system will present a state of mixed driving of ICVs and human-driven vehicles (HDVs) for a long time (i.e., Incomplete Vehicle Networking Environment, IVNE). Due to the different driving behavior of vehicles, the traffic in the ICVs will be more complex, and the intersection will still the bottleneck in the road network.In the ICVs environment, traffic signal control is still the main measures to ensure traffic order and traffic safety at intersections. Traffic signal control of intersections relies on traffic flow models to evaluate the performance of control strategies quickly and accurately. Therefore, this paper established a queue network model with a feedback mechanism for multi-class customers at intersections considering the dynamic randomness of traffic demand, the state-dependence of service, and the congestion propagation among different facilities of intersections simultaneously to accurately describe the performance of the intersection. Furthermore, the theory and method of traffic signal control in the IVNE is studied. The main research work is as follows:(1) The queue network model of intersection with feedback mechanism considering congestion propagation is constructed. To address the inherent randomness in transportation systems, queue systems with random arrival and service processes have been applied. To account for the state-dependent service ability, the state-dependent queue is introduced. To describe the time-varying traffic demand, the fluid queuing theory is introduced. To consider the impact of traffic congestion on traffic demand, the queue with feedback mechanism is introduced. In addition, the facilities of intersection are abstracted into a node queuing system, and the dynamic topology is used to describe the traffic signal control strategy. And then, the queue network model of intersections with a feedback mechanism is constructed to accurately estimate operation performances of the intersection. Simulation results show that the average relative error of the proposed model is 6.4337% under different traffic intensities.(2) Based on the above-mentioned queue network model of intersections with feedback mechanism, a multi-class customers queue network model of intersections with feedback mechanism for IVEN is constructed. By considering the car-following behaviors of ICVs and HDVs, the Fundamental Diagram Model (FDM) of heterogeneous traffic flow is constructed. Further, the relationship between the average speed of heterogeneous traffic flow and traffic density under different penetration of ICVs is studied. And the relationship between the average speed of heterogeneous traffic flow and traffic density is fitted by an exponential speed model to calibrate the service ability of the queue system. Then, different vehicle classes in the IVNE are regarded as different customers, and a queue network model with a feedback mechanism for multi-class customers is constructed. Simulation results show that the average relative error of the proposed model is 6.4463% under different traffic intensities.(3) An optimization method for traffic signal control at intersections considering congestion propagation is constructed. First, the indicators of intersection operation performance such as average delay time, average fuel consumption, and average emissions are deduced. Then, an optimization model aiming at minimizing the average delay time of vehicles is constructed. To adapt to the dynamic randomness of traffic demand, a rolling optimization strategy is designed. The results of numerical experiments show that: 1) The rolling optimization strategy proposed in this paper can effectively reduce the average delay time. 2) With the increase of the penetration rate of ICVs, the optimal average delay time of vehicles is further reduced.(4) An optimization method of traffic signal control at the intersection group considering dynamic traffic assignment is constructed. Considering the interaction between the traffic signal control strategy and the traffic demand, this paper embeds the dynamic traffic assignment algorithm under IVNE in the optimization method of traffic signal control at the intersection group. The path selection behaviors of ICVs and HDVs are described as dynamic user equilibrium (DUE) and Stochastic dynamic user equilibrium (SDUE), respectively. The results show that the optimization algorithm considering dynamic traffic assignment can further reduce the average stay time of vehicles, and can further alleviate the saturation of the node queuing system.