交叉路口是城市中的交通流汇聚场景，传统信控路口中的车辆怠速甚至频繁启停带来了交通效率和燃油经济性的下降。智能网联汽车作为车联网技术与自动驾驶技术的载体，为交通环境的改善提供了可能。现有研究中，多车行为冲突建模与求解方法在计算效率与算法最优性间难以平衡。混合交通研究中，驾驶人驾驶车辆尚未被显式考虑到协同决策控制中，限制了智能网联汽车对交通效率的改善。针对这些问题，本文面向不同智能网联汽车渗透率，提出了多车行为冲突解耦与图论建模方法、混合队列数学建模、性能分析与最优控制框架及面向混合交通的信号灯-车辆协同决策方法，可有效保证交通安全，提升交通效率与燃油经济性。 首先，提出了交叉路口多车行为冲突解耦与图论建模方法。采用有向冲突图与无向共存图构建多车行为冲突模型。由此构建了改进深度优先生成树方法及最小团覆盖方法，分别从局部最优和全局最优的角度求解了多车冲突解耦问题并证明了该问题为NP完全问题。由此提出了对应启发式求解方法，实现了低算法复杂度下的大规模车辆近似最优通行序列求解。构建了换道场景下多车目标分配与无冲突路径规划的迭代求解方法，实现了对多车冲突解耦问题的全解空间建立。 其次，提出了混合队列数学建模、性能分析与最优控制框架。构造了一辆智能网联汽车与$ n $辆驾驶员驾驶车辆组成的“$ 1+n $”混合队列构型。基于线性化车辆动力学模型，导出了构型开环稳定性与可控性的判定条件。基于驾驶员驾驶模型分析，提出了混合交通下混合队列最优速度的计算方法。以此为终端约束，结合车辆动力学模型，构建了混合队列最优控制框架，基于伪谱法对高阶非线性最优控制问题进行求解，保证了智能网联汽车对混合交通整体效率的优化效果。 然后，提出了面向混合交通的信号灯-车辆协同决策方法。针对不同速度轨迹，不同队列规模提出了混合队列收敛时间的分析方法与控制边界可行性判定方法，进而提出了以混合队列与自由驾驶员驾驶车辆为控制单元的混合队列深度优先生成树算法。基于生成树车辆几何构型、通信拓扑结构，车辆动力学模型及安全约束，构建了混合队列最优控制框架，实现了混合交通下的信号灯-车辆协同决策。 最后，搭建了仿真与微缩实车验证平台，分别进行了全网联环境及混合交通环境下的算法功能验证实验。结果表明，所提出的方法可实现交叉路口的多车行为冲突解耦，保证交通安全，综合提升交通效率与燃油经济性。

Intersections are the convergence point in urban traffic scenarios. Traditional signalized intersections lead to vehicle idling and stop-and-go behavior, reducing traffic efficiency and fuel economy. With the emergence of V2X and autonomous technology, connected and automated vehicles (CAVs) provide the possibility to further improve traffic mobility. In the existing research, multi-vehicle behavior conflict solving methods have difficulties in balancing computational efficiency and algorithm optimality. In mixed traffic research, human-driven vehicles (HDVs) have not been explicitly considered in cooperative decision-making and control, which limits traffic efficiency improvement of CAVs. To solve these problems, this thesis proposes a multi-vehicle behavior conflict decoupling and graph-based modeling method, a mixed platoon mathematical modeling, performance analysis, an optimal control framework, and a cooperation method of traffic signal lights and CAVs in the mixed traffic environment. This method can effectively ensure traffic safety and improve traffic efficiency and fuel economy.First, multi-vehicle behavior conflict decoupling and graph-based modeling methods at intersections are proposed. The multi-vehicle behavior conflict model is constructed by using a conflict-directed graph and a coexistence undirected graph. Therefore, the improved depth-first spanning tree method and the minimum clique cover method are constructed, which implies the multi-vehicle conflict decoupling problem is solved from the perspective of local optimum and global optimum respectively, and the problem is proved to be NP-complete. Therefore, a corresponding heuristic solution method is proposed, which solves the approximate optimal CAV passing order for large-scale of vehicles with low algorithm complexity. An iterative solution method for multi-vehicle target assignment and conflict-free path planning in lane-changing scenarios is constructed, and complete solution space for the multi-vehicle conflict decoupling problem is established.Secondly, the mathematical modeling, performance analysis, and optimal control framework of the mixed platoon are proposed. A ``1+n‘‘ mixed platoon structure consisting of a leading CAV and n following HDVs is proposed. Based on the linearized vehicle dynamics model, the open-loop stability and controllability conditions of the mixed platoon structure are derived. Based on the analysis of the car following model, the calculation method for the optimal speed of mixed platoon is also proposed. Taking the optimal speed as the terminal constraint, combined with the vehicle dynamic model, the mixed platoon optimal control framework is constructed, and this high-order nonlinear optimal control problem is solved based on the pseudo-spectral method, which ensures the CAVs’ optimization effects on the overall efficiency in the mixed traffic environment.Then, a traffic signal light and CAV cooperation decision-making method in the mixed traffic environment is proposed. For different speed trajectories and different platoon sizes, the convergence time analysis method of the mixed platoon and the determining method for corresponding control feasibility boundary are proposed. A mixed platoon depth-first spanning tree algorithm is further proposed with the mixed platoon and the free-driving HDVs as the control unit. Based on the spanning tree geometry, communication topology, vehicle dynamic model, and safety constraints, an optimal control framework for mixed platoon is constructed to realize traffic signal light and CAV cooperation decision-making under mixed traffic.Finally, the simulation and miniature vehicle sandbox verification platforms are built, and the algorithm verification experiments in the full CAV penetration rates environment and the mixed traffic environment are carried out respectively. The results show that the proposed method can realize the decoupling of multi-vehicle behavior conflicts at intersections, ensure traffic safety, and comprehensively improve traffic efficiency and fuel economy.