智能网联环境下面向城市交通网络的多车协同预测控制方法研究

In view of the problem that the existing coordinated control methods for multiple vehicles are limited to certain scenarios, the state optimization for intelligent and connected vehicles (ICVs) in urban traffic network will be studied and a hierarchical controller will be designed. In the higher level controller, the traffic system will be modeled based on the network topology, and in the lower level controller, a decentralized optimization of the vehicle states will be performed. Due to the fact that the existing optimization models are not universal, coordinated optimization of multiple vehicles based on robust model predictive control considering uncertainties will be researched. Based on the vehicle state information obtained from vehicle to vehicle and vehicle to infrastructure communications, a closed-loop constrained nonlinear optimization model with bounded uncertainties will be built, in which the cost function consists of the fuel efficiency and comprehensive comfort, and the constraints include the safety and traffic mobility. The optimization objects in the cost function will be decoupled, the critical stability condition of the control system will be deduced, and the online optimization solution of the algorithm will be discussed. A driver-in-the-loop coordinated optimization problem for multiple vehicles based on stochastic model predictive control will be modeled, and the simplification method of the model as well as the fast computation algorithm will be researched. A method for the evaluation of the energy consumption of ICVs will be designed based on orthogonal experiment and neural network. The control methods will be validated using software simulation, hardware-in-the-loop experiment and vehicle-in-the-loop experiment. This proposal is aimed at achieving the implementation of the coordinated predictive control algorithms for multiple vehicles in engineering practice, and providing theoretical support for the industrialization applications of the ICV technology.

针对现有的多车协同控制方法局限于特定场景的问题，研究智能网联环境下面向城市交通网络的车辆状态分层优化。设计分层控制器，上层基于网络拓扑构架对交通系统进行建模，下层对车辆状态进行分布式优化。由于现有的优化模型不具备普适性，研究考虑不确定性时基于鲁棒模型预测控制的多车协同优化。基于V2V和V2I获取的车辆状态信息，以燃油经济性和综合舒适性为优化目标，以安全性和交通流畅性为约束条件，建立考虑有界不确定性的闭环约束非线性优化模型。对优化模型中的优化目标进行解耦，推导出系统临界稳定性条件，研究算法的在线优化。研究驾驶员在环时基于随机模型预测控制的多车协同优化问题的建模、简化及快速求解算法。采用正交试验和神经网络，设计智能网联汽车能耗评价方法。基于软件仿真、硬件在环试验和实车在环试验，对控制算法进行验证。本课题旨在实现面向工程实际的智能网联多车实时协同预测控制，为智能网联技术的产业化应用提供理论支持。

基于智能网联技术的多车协同控制可以有效提升城市交通的通行效率和车辆的燃油经济性。本项目以智能网联汽车为研究对象，对城市信号交叉口场景下的多车协同控制方法进行了深入研究，主要研究内容及成果如下：（1）设计了信号交叉口的分层控制架构，提出一种信号灯与智能网联汽车分层协同控制方法，上层为优化交通信号配时的交通信号控制器，下层为优化车辆速度轨迹的多车协同控制器，仿真验证了分层协同控制方法的有效性；提出一种混合交通环境下的多车协同控制方法，将智能网联汽车与传统人类驾驶汽车组成队列，通过优化智能网联汽车的速度轨迹降低整个车队的能量消耗。（2）考虑实际通信环境下时延导致的不确定性，提出了一种基于鲁棒模型预测控制的多车协同控制方法，对协同控制系统的稳定性进行理论分析，有效保障了车队轨迹的稳定跟踪。（3）基于实车试验数据建立了驾驶员误差预测模型，提出了一种基于随机模型预测控制的网联汽车多车协同控制方法，有效降低了驾驶员误差导致的车辆速度轨迹偏移；提出了一种随机模型预测控制算法的快速求解方法，提高多车协同控制算法的实时性。（4）基于正交试验表进行仿真试验，分析了智能网联汽车能耗的影响因素，建立了能耗与影响因素的数学模型，设计了能耗评价表。（5）基于智能网联小车试验平台和智能网联试验车对多车协同控制方法进行硬件在环和实车在环试验。本项目的研究成果可以为后续智能网联汽车多车协同控制的研究和实际应用提供理论参考。