【作者】 崔欣；

【导师】 侯忠生；

【作者基本信息】 北京交通大学 ， 控制理论与控制工程， 2015， 硕士

【摘要】 多智能体系统和无模型自适应控制方法得到越来越广泛的关注,前者是分布式人工智能的重要分支,后者则是典型的数据驱动控制方法,二者在解决复杂系统相关问题时也有着各自的优势。本文目的于实现多智能体系统理论和无模型自适应控制方法的有机结合,并在多交叉口信号灯的协调控制中加以应用。主要的研究内容如下：首先,构建了多智能体系统协调矩阵的表达方法,并基于多入多出离散时间系统的无模型自适应控制理论,对多智能体系统,设计了一种无模型自适应协调控制算法(Model Free Adaptive Coordinated Control, MFACC)。仿真分析结果表明所提出的无模型自适应协调控制算法能够完成多智能体的独立跟踪、协调跟踪以及两者相结合的控制任务。其次,针对道路交通控制中的重点研究对象之一——多交叉口信号灯的协调控制,讨论了对其应用多智能体系统理论的意义和可行性。然后,根据排队长度均衡思想,把传统的无模型自适应控制和所提出的无模型自适应协调控制分别应用于多交叉口排队长度均衡的信号灯控制设计中。仿真结果表明,两种无模型自适应控制方法都能够达到交叉口排队长度均衡的控制目标,且无模型自适应协调控制不仅能实现单交叉口的排队长度均衡,还能实现交叉口之间的排队长度实现均衡,即多智能体系统的协调。最后,考虑到多智能体系统间的互联影响会对多智能体协调控制效果产生的影响,以多交叉口智能体系统为例,对多智能体系统的模型进行了改进,分别提出了改进的交叉口存储转发模型和基于交通波理论的关联交叉口排队长度模型,并运用复杂系统的无模型控制方法对系统进行分解,设计了对多交叉口智能体系统的分散估计分散控制型MFAC (Model Free Adaptive Control)算法。仿真结果表明,针对多交叉口系统的改进多智能体模型更加符合实际交通状况,尤其是当主干道交通流密度很大时,分散估计分散控制型MFAC算法能大大提升均衡控制的效果。更多还原

【Abstract】 Multi-agent systems and model free adaptive control (MFAC) scheme have been gotten more and more attentions nowadays. The former is an important branch of distributed artificial intelligence, the latter is the typical data-driven control. They all have the advantages in solving problems related to complex systems. This thesis is to combine both techniques and apply to the coordinated control of intersections. The main research contents are as follows:Firstly, the methods of multi-agent systems’ coordination matrix are built, and based on multiple-input-multiple-output discrete-time systems’model free adaptive control theory, the model free adaptive coordinated control (MFACC) scheme is designed for multi-agent systems. The simulation results show that the proposed model free adaptive coordinated control scheme can complete the control task of the Multi-agent systems’ single tracking, coordinated tracking or the combination.Secondly, for the signal coordinated control of multiple intersections, which is the main research subject of the road traffic control, the importance and feasibility of combining multi-agent systems and road traffic control are discussed. Then, according to the idea of equilibrium queue length, the traditional model free adaptive control and the proposed model free adaptive coordinated control are applied to design the intersections’ queue length balanced signal controller. The simulation results show that both schemes can achieve the target of vehicle queuing length balanced-control, and model free adaptive coordinated control not only can meet the target of queuing length balanced-control for single intersection, but also can make the equilibrium of the queuing length of the intersections, that is, multi-agent systems’ coordination.Finally, considering the mutual influence among multi-agent systems can impact the multi-agent systems’ coordination effect, taking the intersection multi-agent systems for example, aiming to improve the multi-agent systems’ model, and the improved store-and-forward modeling and the queue length model of adjacent signal intersections which is based on traffic-wave theory have been presented. Using the complex systems’ model free adaptive control scheme to compose of the systems, so that we can develop decentralized control based on decentralized estimation of multi-agent systems faced to intersections. The simulation results show that the improved multi-agent models faced to intersections are more closely with the practical traffic conditions, especially when the main road traffic density is very large, decentralized control based on decentralized estimation MFAC strategy improves balanced-control performance greatly.更多还原