返回
类型 综合研究 预答辩日期 2017-09-18
开始(开题)日期 2015-09-10 论文结束日期 2017-06-08
地点 东南大学数学学院第一报告厅 论文选题来源 国家自然科学基金项目     论文字数 8.5 (万字)
题目 时滞网络系统的动力学分析及采样控制
主题词 时滞网络系统,多智能体系统,时滞,同步,采样控制
摘要 神经网络,多智能体系统,复杂网络信息物理系统等多种时滞网络系统的动力学分 析和一致性控制是今年来的热点问题,其在通信安全,神经科学,飞行器编队,大规模传 感器网络,交通网络等众多领域有着广泛应用,引起了不同领域学者的密切关注。研究 这些时滞网络中的动力学和同步现象,将有利于人们更深刻地理解网络环境下的协同机 理,也有助于设计更有效的控制协议以实现多样的控制目标。本文主要基于Lyapunov稳 定性理论,矩阵测度方法,采样控制,网络安全控制思想等,对多种时滞网络系统的动 力学,同步以及分布式一致性控制等问题进行了探讨。全文由以下七个部分组成: 第一章首先介绍了时滞动态网络系统的背景和研究意义,具体阐述了时滞神经网 络的研究进展和待解决的热点问题,此外,介绍了多智能体系统协同控制以及采样控制 的研究意义和研究进展,并在此基础上详细阐述了本文的主要研究内容和主要创新点。 第二章主要研究了两类时滞神经网络的动力学和同步控制问题。第一节讨论了带 有时变时滞的惯性神经网络的稳定性和同步控制。利用矩阵测度和Halanay不等式,我 们得到了保证惯性神经网络平衡点全局指数稳定的充分性判据。所得的判据以矩阵测 度形式给出,判据含义直观且易于验证。通过设计线性反馈控制,得到了可以保证主从 惯性神经网络同步,控制器增益需满足的充分性条件。第二节讨论了带有参数不确定性 和时变时滞的主从Cohen-Grossberg神经网络的固定时间同步问题。这里的固定时间同步 概念中,停息时间与初值无关,故而可以事先人为设定和调整。我们提出了新的固定时 间同步控制策略。利用Filippov不连续理论,得到了保证固定时间同步的控制器参数选 取方法。 第三章讨论了几类动态网络的采样控制问题。第一节中,研究了有向拓扑和随机采 样机制下,双层多智能体系统的均方点到点一致性跟踪问题。通过输入延迟方法和构 造不连续Lyapunov泛函,得到了保证每个跟随者的状态在均方意义下渐近跟踪到相应领 导者状态的充分性条件。在第二节,考虑了有向图下,多智能体系统基于观测器的镇定 问题,并设计了基于采样信息的分布式镇定控制协议。为了镇定整个网络中智能体的状 i态,除了所有智能体均可以利用采样时刻的相对输出信息外,假定只有一小部分节点可 以利用自身的绝对输出信息。基于输入到状态稳定的定义和性质以及Lyapunov稳定性理 论,我们给出了控制器增益,观测器增益,耦合强度的设计方法,以及可容许采样间隔 需满足的条件。此外,还讨论了相应线性矩阵不等式判据的可解性。在第三节,研究了 带有分布式传感器,量化过程,和网络通讯时滞的主从混沌神经网络的同步问题。在每 个采样时刻,根据指定的调度策略,仅选取其中一个传感器将其最新更新的状态传送至 控制器端。从而,基于调度的通讯和控制策略将更为节能。为了保证主从混沌系统在调 度协议下的量化同步,我们得到了关于输出反馈增益矩阵,可容许的采样间隔,以及网 络时延上界的充分条件。 第四章研究了间歇控制下,带有线性动力学的多智能体系统的分布式鲁棒镇定问 题。在有向图下,假设每个智能体的动力学中带有未知不确定性。假设根节点的控制输 入可以额外地间歇利用自身的绝对状态信息,而其他所有节点可以连续地利用邻居的 相对状态信息。为了镇定整个网络的状态,我们给出了选取控制参数的算法,并利用有 向图理论和Lyapunov稳定性理论给出了间歇控制的时间比率。 第五章在信息物理系统的框架下,讨论了带有Lipschitz非线性的动态网络系统的分 布式跟踪问题。由于实际中的诸多限制,智能体的状态往往不可直接获取并用来设计控 制器,故而首先设计了观测器重构智能体的状态信息。假设控制器和观测器的通讯网络 可能遭受序列的网络攻击,这些攻击将破坏相应通讯拓扑的连通性且攻击对不同的通 讯网络的影响可以不同且相互独立。给出了新的网络安全控制协议,并提供了选取反馈 增益矩阵和耦合强度的算法。 第六章考虑了多区域时变时滞下城市路网的自适应周界控制问题。首先,基于宏观 基本图概念,我们提出了区域车辆数所满足的非线性常微分方程模型。该模型包含车辆 驶往区域边界的行驶时间以及交通拥塞的疏散过程,分别建模为输入和状态时滞。控制 目标为调节每个区域的车辆数至期望值。通过利用模型参考自适应控制和渐近滑模控 制方法,仅利用参考模型的信息,我们设计了相应的自适应控制策略,并证明了跟踪误 差的稳定性。 第七章对该论文工作进行了全面的总结,并对今后的研究方向进行了展望。
英文题目 Analysis of the Dynamics in Delayed Networked System and Sampled-Data Control
英文主题词 Delayed networked system, Multi-agent systems, Time delay, Synchronization, Sampled-data control
英文摘要 The analysis and synthesize of dynamics as well as distributed consensus control of delayed networked system are becoming hot research topics in recent years, including neural networks, multi-agent systems, complex cyber-physical systems, etc. Many researchers from different fields have paid much attention to it’s wide applications in communication security, neuroscience, spacecraft formation flying, large-scale sensor networks, urban traffic networks, etc. Investigations of dynamics and consensus phenomena of delayed networks will contribute to a deep comprehension of cooperative mechanisms in networked circumstances, and will also help to design more effective protocols to realize various control objectives. Mainly based on Lyapunov stability analysis, matrix measure approach, sampled-data control, cyber-security control, this dissertation is devoted to analyze the dynamics, synchronization and distributed control of several delayed networked systems. Specifically, this dissertation is divided into seven chapters and organized as follows: In the first chapter, the general backgrounds and the significance of delayed networked system are presented. Specifically, the developments and hot topics of the delayed neural networks are elaborated. Furthermore, the significance and developments of cooperative control of multi-agent system and sampled-data control are stated. The main contents and contributions of this dissertation are explained based on the above-mentioned discussions. In the second chapter, the dynamics and synchronization control of two kinds of delayed neural networks are explored. In Section 2.1, the stability and synchronization control of an inertial neural network with time-varying delays is concerned. By using matrix measure and Halanay inequality, several sufficient conditions for global exponential stability of the equilibrium are provided in form of matrix measure. These criteria are simple in form and easy to verify. To realize the synchronization of master-slave neural networks, an error-feedback control strategy is employed and requirements for the feedback gains are also derived. In iiiSection 2.2, the fixed-time master-slave synchronization of Cohen-Grossberg neural networks with parameter uncertainties and time-varying delays is investigated. The settling time of fixed-time stability can be adjusted to desired values regardless of initial conditions. Novel synchronization control strategy is proposed. By utilizing Filippov discontinuous theory and Lyapunov stability theory, some sufficient conditions are provided for selecting the control parameters to ensure synchronization with required convergence time. In the third chapter, several dynamical networks with sampled-data control are discussed. Section 3.1 is concerned with the mean square node-to-node consensus tracking problem for multi-agent systems with stochastic sampling and directed graphs. By employing the inputdelay method and constructing discontinuous Lyapunov functionals, it arrives at sufficient conditions under which the state of each follower can track that of the corresponding leader asymptotically in the mean square sense. In Section 3.2, the distributed observer-based stabilization problem of multi-agent systems under a directed graph is investigated and distributed control protocol with sampled-data information is proposed. In order to stabilize the states of the whole network, all the nodes can utilize the relative output estimation error at sampling instants and only a small fraction of nodes can use the absolute output estimation error additionally. By virtue of input-to-state stability and Lyapunov stability theory, an algorithm to design the control gain matrix, observer gain matrix, coupling strength as well as the allowable sampling period are derived. Further discussions about the solvability of obtained linear matrix inequalities are also given. In Section 3.3, the synchronization problem of master-slave chaotic neural networks with distributed sensors, quantization process, and communication delays is investigated. At each sampling instant, only one sensor is scheduled to transmit its latest information to the controller’s side. Thus, such communication and control strategy are much more energy-saving. Sufficient conditions for output feedback control gain matrix, allowable length of sampling intervals, and upper bound of network-induced delays are derived to ensure the quantized synchronization of master-slave chaotic neural networks. The distributed robust stabilization control problem of multi-agent systems with linear dynamics is investigated in the fourth chapter. The considered topology is directed and the dynamics of each agent are subjected to unknown uncertainties. Relative state feedback control ivinputs are implemented for every node while additionally, the control input of the root agent in a spanning tree can utilize its own absolute state intermittently. In order to stabilize the whole network, an algorithm to choose control parameters is provided and the required length of the intermittent control intervals is also derived by using directed graph theory and Lyapunov stability analysis. In the fifth chapter, under the framework of cyber-physical systems, distributed tracking problem for complex dynamical networks with Lipschitz-type nonlinear dynamics is investigated. Due to limitations in practical circumstances, the states of the agents are usually unavailable for controllers, so distributed observers is firstly designed to reconstruct the states of nodes. The communication channels for controllers and observers may subjected to a sequence of malicious attacks, which will destroy the connectivity of the communication topology. It is assumed that the impacts of attacks on different communication networks can be different and independent. New security control strategies are proposed and analyzed. An algorithm to properly select the feedback gain matrices and coupling strengths is also presented. In the sixth chapter, an adaptive perimeter control problem is studied for urban traffic networks with multiple regions and time-varying delays. Firstly, system model is formulated as nonlinear ordinary differential equations based on the concept of macroscopic fundamental diagram. This model includes both the travel times of vehicles traveling to the borders of regions as well as evacuation process of traffic jams, and they are modeled as state and input delays, respectively. The control objective is to stabilize the number of vehicles in each region to desired values. By employing the model reference adaptive control combining with asymptotical sliding mode technique, adaptive laws for control parameters are given by using only the information of the reference model. Finally, the stability of tracking error is analyzed. In the seventh chapter, the research work of this paper is summarized, and some interesting future researches are also included.
学术讨论
主办单位时间地点报告人报告主题
东南大学复杂系统与网络科学研究中心 2015.3.17 东南大学数学学院第二报告厅 万颖 Decentralized robust adaptive control for the multi-agent system consensus problem using neural networks
东南大学复杂系统与网络科学研究中心 2015.10.23 东南大学数学学院第二报告厅 万颖 adaptive control of uncertain nonaffine nonlinear systems with input saturation using neural networks
东南大学复杂系统与网络科学研究中心 2016.4.5 东南大学数学学院第二报告厅 时欣利 Asymptotic convergence of constrained primal-dual dynamics
东南大学复杂系统与网络科学研究中心 2015.5.14 东南大学数学学院第二报告厅 万颖 A continuous asymptotic tracking control strategy for uncertain nonlinear systems
东南大学复杂系统与网络科学研究中心 2015.9.15 东南大学数学学院第二报告厅 万颖 A Round-Robin type protocol for distributed estimation with H∞ consensus
东南大学复杂系统与网络科学研究中心 2016.4.12 东南大学数学学院第二报告厅 李若霞 Predictor-based sampled-data exponential stabilization through continuous–discrete observers
东南大学复杂系统与网络科学研究中心 2016.5.17 东南大学数学学院第二报告厅 黄群 Leader-following consensus of a class of stochastic delayed multi-agent systems with partial mixed impulses
东南大学复杂系统与网络科学研究中心 2016.5.19 东南大学数学学院第二报告厅 王毅 Frozen state conditions for exponential consensus of time-varying cooperative nonlinear networks
     
学术会议
会议名称时间地点本人报告本人报告题目
第十届亚洲控制会议 2015.5.31-6.3 哥打基纳巴卢,马来西亚 Distributed robust control of linear multi-agent systems under directed topology
第28届中国控制与决策会议(2016CCDC) 2016.5.27-5.31 宁夏,中国 Distributed H∞ node-to-node consensus via aperiodic sampled-data pinning control
     
代表作
论文名称
Matrix measure strategies for stability and synchronization of inertial BAM neural network with time
Distributed node-to-node consensus of multi-agent systems with stochastic sampling
Periodicity and synchronization of coupled memristive neural networks with supremums
Distributed robust stabilization of linear multi-agent systems with intermittent control
Robust fixed-time synchronization of delayed Cohen-Grossberg neural networks
Distributed observer-based stabilization of nonlinear multi-agent systems with sampled-data control
Distributed node-to-node consensus via aperiodic sampled-data pinning control
 
答辩委员会组成信息
姓名职称导师类别工作单位是否主席备注
赵洪涌 正高 教授 博导 南京航空航天大学
肖敏 正高 教授 博导 南京邮电大学
梁金玲 正高 教授 博导 东南大学
卢剑权 正高 教授 博导 东南大学
虞文武 正高 教授 博导 东南大学
      
答辩秘书信息
姓名职称工作单位备注
聂小兵 副高 副教授 东南大学