异构车辆队列的协同预测控制研究

As one of the hot topics in the field of intelligent transportation systems, the platooning of connected and automated vehicles can usually be solved by regarding it as the multi-agent consensus problem. However, the particularity of vehicle models and the complexity of traffic environment bring platoon control many challenges, such as the nonlinearity and heterogeneity of vehicle models, system constraints and external disturbances. Existing research indicates that model predictive control is capable of tackling these issues. Therefore, this project is to study the cooperative predictive control of heterogeneous vehicle platoons with system constraints and general communication topologies, by exploring the cooperative predictive control based consensus theory and algorithms. Firstly, we propose a cooperative predictive algorithm for heterogeneous multi-agent consensus, and provide a theoretical framework of its asymptotic consensus and robustness. Secondly, we propose a novel distributed control algorithm for heterogeneous vehicle platoons based on cooperative predictive consensus, and provide internal stability and string stability analyses of the vehicle platoons. Finally, we realize the platoon control of intelligent mobile vehicles to verity the theoretical results. This project will deepen the coordination theory of multi-agent systems, and provide novel cooperative predictive control theory and algorithms for heterogeneous vehicle platoons to improve its robustness and adaptability, so as to promote the development and application of cooperative automatic driving technique.

作为智能交通系统研究的热点之一，车辆队列控制可以抽象为多智能体一致性问题来解决，而车辆模型的特殊性以及交通环境的复杂性给车辆队列控制问题带来模型非线性且异构、系统约束和外界干扰等诸多挑战。现有研究表明模型预测控制在应对上述挑战时具有其先天的优势。因此，本项目将模型预测控制与多智能体协同理论相结合，通过发展协同预测一致性理论与方法来研究一般通信拓扑下存在系统约束的异构车辆队列控制问题。首先，提出异构多智能体一致性的协同预测控制算法，建立其渐近一致性和鲁棒性的理论框架；然后，提出基于协同预测一致性的异构车辆队列控制算法，对车辆队列系统的内稳定性和队列稳定性进行分析；最后，在智能小车平台上对车辆队列控制理论结果进行试验验证。本项目研究将在深化多智能体协同理论的同时，为异构车辆队列控制提供新的协同预测控制理论与方法，以改善车辆队列系统的鲁棒性和可扩展性，进而推动协同式自动驾驶技术的发展及其应用。

作为智能交通系统研究的热点之一，车辆队列控制可以抽象为多智能体一致性问题来解决，而车辆模型的特殊性以及交通环境的复杂性给车辆队列控制问题带来模型非线性且异构、系统约束和外界干扰等诸多挑战。现有研究表明模型预测控制在应对上述挑战时具有其先天的优势。因此，本项目将模型预测控制与多智能体协同理论相结合，通过发展协同预测一致性理论与方法来研究一般通信拓扑下存在系统约束的异构车辆队列控制问题。首先，我们提出了基于分布式模型预测控制的事件触发/自触发一致性协议，建立其渐近一致性理论框架；针对带扰动和/或时延、切换通信拓扑的多智能体系统提出了一系列鲁棒一致性协议，并给出了相应的鲁棒一致性判据；针对异构多智能体系统提出了一致性协议和渐近一致性判据。然后，提出了一种基于分布式模型预测控制的异构车辆队列控制算法，证明了所提出算法的迭代可行性以及队列的内稳定性和队列稳定性；针对具有任意通信时滞和采样间隔的异步通信车辆队列提出了一致性控制协议，并在通信时滞和采样间隔任意大条件下证明了车辆队列的内稳定性和队列稳定性。最后，在智能小车平台上对车辆队列控制理论结果进行了实验验证。通过本项目，我们发展了多智能体一致性理论，为异构车辆队列控制提供了新的协同预测控制理论与方法，改善了车辆队列系统的鲁棒性和适用性。