【作者】 谭艳；

【导师】 边有钢；

【作者基本信息】 湖南大学 ， 机械工程， 2023， 硕士

【摘要】 智能化、网联化作为汽车行业的发展趋势,为网联多车协同控制提供了技术支持。本文以智能网联车辆队列为研究对象,基于分布式模型预测控制(Distributed Model Predictive Control,DMPC)方法,针对燃油经济性及随机通信丢包问题分别构建了车辆队列预测控制架构,设计了分布式模型预测控制器,导出并证明了保证车辆队列内稳定性的具体设计条件,最后通过仿真验证并分析了所设计的控制器的综合性能。全文的工作如下:(1)构建了网联车辆队列的系统架构与模型。首先,建立了状态空间形式的车辆节点纵向动力学模型。其次,基于图论定义了前车-领航者跟随式以及双前车跟随式两种信息流拓扑结构。接着,针对以燃油经济性为导向的优化问题建立了功率多项式形式的油耗模型;针对随机通信丢包分别建立了伯努利丢包模型以及马尔可夫丢包模型。最后,基于恒定间距策略给出了队列的跟踪控制目标,为解决燃油经济性优化与随机通信丢包问题提供了模型依据。(2)设计了考虑燃油经济性的分布式模型预测控制器。以跟踪领航车、维持安全间距、降低油耗为目标,构建了双层分布式模型预测控制架构。上层控制器以跟踪控制为目标得到最优跟踪控制输入,下层控制器以降低油耗为目标得到最优经济控制输入,其中,两层控制器通过约束条件紧密联系。最后基于李雅普诺夫稳定性理论分析并证明了系统的渐近稳定性。(3)设计了考虑通信随机丢包的分布式模型预测控制器。以跟踪控制为目标分别构建了伯努利丢包、马尔可夫丢包下的分布式模型预测控制架构。引入了均方渐近稳定的概念,导出了两种丢包模型下车辆队列实现均方渐近稳定的条件。(4)仿真验证了考虑燃油经济性与随机通信丢包问题的控制器的性能。针对以经济性为导向的分布式控制器,在正弦扰动工况下进行数值仿真,通过与基准方案进行对比,验证了所设计的两个改进方案对综合性能的改善。针对考虑随机通信丢包的分布式控制器,在加速扰动工况下分别对两种丢包模型进行了仿真验证,分析了丢包模型参数对控制性能的影响。更多还原

【Abstract】 As the development trend of the automobile industry,intelligence and network connectivity have provided the technical support for the networked multi-vehicle cooperative control.In this paper,relying on the Distributed Model Predictive Control(DMPC)method,a vehicle platoon predictive control architecture is constructed for the fuel economy and the random communication packet loss problems,besides,the DMPC controllers are designed,then specific design conditions to guarantee the internal stability of the vehicle platoon are proposed and demonstrated.Finally,the comprehensive performance of the designed controllers is verified and analyzed through simulations.The work of the full paper is as follows:(1)The system architecture and the model of the networked vehicle platoon are constructed.Firstly,a longitudinal dynamics model of vehicle nodes in the form of state space is established.Secondly,the Predecessor-Leader Following and the TwoPredecessor Following information topologies are defined based on graph theory.Then,the fuel consumption model in the form of power polynomial is developed for the fuel economy-oriented optimization problem;the Bernoulli packet loss model and the Markov packet loss model are developed for random communication packet loss,respectively.Finally,the tracking control objective of the platoon is given based on the constant spacing strategy,which provides a model basis for addressing the multiobjective optimization and the random communication packet loss problems.(2)A distributed model predictive controller considering fuel economy is designed.A double-layer distributed model predictive control architecture is constructed with the objectives of tracking the leading vehicle,maintaining safe spacing,and reducing fuel consumption.The upper layer controller acquires the optimal tracking control input with the goal of tracking control,and the lower layer controller acquires the optimal economic control input with the goal of fuel consumption reduction,where the two layers are closely linked by constraints.Finally,the asymptotic stability of the system is analyzed based on the Lyapunov stability theory.(3)A distributed model predictive controller considering random communication packet loss is designed.The distributed model predictive control architectures under the Bernoulli packet loss and Markov packet loss are constructed with the objective of tracking the leading vehicle respectively.The concept of mean-square asymptotic stability is introduced,and the conditions for achieving this stability of the vehicle platoon under the two packet loss models are derived.(4)The performance of the controllers considering fuel economy and random communication packet loss issues is verified by simulations.For the economy-oriented distributed controller,numerical simulations are carried out under the sinusoidal perturbation scenario,and the improvement of the comprehensive capability of the two designed improvement schemes is verified by contrasting them with the benchmark solution.For the distributed controller considering random communication packet loss,two packet loss models are simulated under the accelerated perturbation scenario to investigate the effect of the packet loss model parameters on the control performance.更多还原