基于分布式模型预测控制的城市快速路多匝道协同控制  

Coordinated Ramp Metering for Urban Expressway Based on Distributed Model Predictive Control

作　　者：杨雪驰 

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海

出　　处：《建模与仿真》2023年第4期3550-3563,共14页Modeling and Simulation

摘　　要：模型预测控制(Model Predictive Control, MPC)是一种适用于多匝道协同控制的控制策略,但其因计算时间过长而难以在复杂路网中实时实施。为了减少MPC控制复杂路网的计算时间,采用分布式控制思想,考虑交通流的单向性,提出一种贯序分布式多匝道MPC控制方案,各控制器依次根据上游控制器传递的交通信息确定最优匝道控制率。最优控制模型采用METANET交通流模型作为过程模型,综合考虑通行效率、匝道队列和控制信号波动构建目标函数,通过求解目标函数在线优化问题得到最佳的入口匝道控制率。仿真结果表明,相较于集中式MPC控制和分散式MPC控制,所提出的贯序分布式多匝道MPC控制方案可以在提高路网表现和降低计算复杂度间达到平衡。Model Predictive Control (MPC) is a control strategy suitable for cooperative control of multiple ramps, but it is difficult to implement in real time in complex road networks due to the long com-putational time. In order to reduce the computational time for MPC control of complex road net-works, a sequential distributed multi-ramp MPC control scheme was proposed, taking into account the unidirectional nature of traffic flow, where each controller determined the optimal ramp me-tering rate in turn based on the traffic information transmitted by the upstream controller. METANET traffic flow model was adopted as the process model, and the objective function was con-structed by considering traffic efficiency, ramp queue and control signal fluctuation. The simulation results showed that the proposed sequential distributed MPC ramp metering strategy offered a trade-off between computational complexity and system performance compared to the centralized MPC strategy and decentralized MPC strategy.

关 键 词：入口匝道控制 交通流模型 通行效率 模型预测控制 城市快速路 最优控制模型 分布式控制 在线优化 

分 类 号：U41[交通运输工程—道路与铁道工程]