城市交通区域的迭代学习边界控制

发布时间：2018-03-12 11:58

本文选题：边界控制　切入点：宏观基本图　出处：《北京交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着汽车保有量的急速增长,许多大城市面临着区域性的拥堵问题,对社会生活和国民经济都造成了严重影响。因此,城市交通系统优化控制是智能交通领域的重要科学问题和重要研究方向。近些年,许多国内外学者开始从宏观层面利用宏观基本图(Macroscopic Fundamental Diagram,MFD)来设计边界控制器以调节整个路网的车流量,使路网通行能力最大化。目前,已有基于MFD的控制方法多属于基于模型的反馈控制算法。然而,实际路网的MFD模型参数难于标定且易受环境影响发生变化,进而影响这些控制算法的实际应用效果。随着信息、传感、计算机等技术的快速发展,城市交通系统每时每刻都在采集和存储蕴含交通系统运行状态的海量数据。此外,城市交通系统的交通流按天、按周、按季度甚至按年的运行模式都具有重复性。迭代学习控制(Iterative Learning Control,ILC)充分利用系统重复运行信息构造控制器,使系统控制性能随运行次数的增加不断得到改善。因此,基于城市交通流的重复特性,本文提出了城市交通区域的两种迭代学习边界控制方案,给出了相应的误差收敛性分析,并通过大量仿真验证了迭代学习控制方法对于各种场景下的区域路网交通情况均能达到较为理想的控制效果。论文主要工作如下:第一,介绍了 MFD产生的背景、基本性质、影响因素及相关参数的计算公式。给出了基于MFD的城市交通系统模型并作了解析。第二,借鉴宏观基本图和迭代学习控制的现有研究成果,提出了城市交通区域的基于MFD的迭代学习边界控制方案,给出了误差收敛性分析,从宏观层面对不同路网情况(早晚高峰、中心区域拥堵、交通需求时变、不同的期望车辆数、放缩MFD)进行了仿真分析,结果显示,当系统参数仅在时间轴上发生变化时,迭代学习边界控制方案对于各种场景下的区域路网交通状态具有良好的控制效果。第三,为抑制交通系统参数的迭代变化,进一步提出了城市交通区域的前馈反馈迭代学习控制方案、给出误差收敛性证明,并从不同路网情况(早晚高峰、中心区域拥堵、不同的期望车辆数)与迭代学习边界控制方案进行了仿真比较,结果表明,前馈反馈迭代学习控制器在系统参数沿时间轴和迭代轴均发生变化时,仍可以实现误差的快速收敛,保证高精度跟踪的同时具有更好的抗扰动能力。
[Abstract]:With the rapid growth of car ownership, many big cities are facing regional congestion problems, which have a serious impact on social life and national economy. The optimization control of urban traffic system is an important scientific problem and important research direction in the field of intelligent transportation. In recent years, Many scholars at home and abroad have begun to design boundary controllers from the macro level using Macroscopic Fundamental Diagram MFDs to regulate the traffic flow of the entire road network and maximize the capacity of the network. Most of the control methods based on MFD are model-based feedback control algorithms. However, the parameters of the MFD model of the actual road network are difficult to calibrate and easily changed by the environment, which will affect the practical application effect of these control algorithms. With the rapid development of sensing, computer and other technologies, urban traffic systems are collecting and storing huge amounts of data containing the state of traffic systems at every moment. In addition, traffic flows in urban transportation systems are daily and weekly. The iterative learning control (iterative Learning control) makes full use of the repeated operation information of the system to construct the controller, which makes the control performance of the system improve with the increase of running times. Based on the repeated characteristics of urban traffic flow, two iterative learning boundary control schemes for urban traffic area are proposed, and the corresponding error convergence analysis is given. Through a large number of simulations, it is verified that the iterative learning control method can achieve a better control effect on the traffic situation of the regional road network under various scenarios. The main work of this paper is as follows: firstly, the background and basic properties of MFD are introduced. The model of urban traffic system based on MFD is given and analyzed. Secondly, the existing research results of macroscopic basic map and iterative learning control are used for reference. In this paper, an iterative learning boundary control scheme based on MFD for urban traffic area is proposed, and the error convergence analysis is given. At the macro level, different road network conditions (morning and evening peak, congestion in central area, time-varying traffic demand, different expected number of vehicles) are analyzed. The simulation results show that when the system parameters change only on the time axis, the iterative learning boundary control scheme has a good control effect on the traffic state of the regional road network under various scenarios. In order to restrain the iterative variation of traffic system parameters, a feedforward and feedback iterative learning control scheme for urban traffic area is proposed, and the error convergence proof is given. The simulation results show that the feedforward feedback iterative learning controller can realize the fast convergence of the error when the parameters of the system change along the time axis and the iterative axis. At the same time, it has better anti-disturbance ability.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491

【参考文献】