多智能体系统（给定）有限时间控制问题研究
发布时间：2018-10-24 07:23
【摘要】：大规模化网络系统具有复杂动特性,其控制问题既十分重要又极具挑战性。传统的集中式控制(Centralized Control)方法已经很难满足复杂大系统实现协调、稳定、高效运行的要求。不同于传统的集中式控制方法,分布式控制(Distributd Control)方法具有所用信息量少,协作性好,灵活性高,可扩展性强等诸多优点,旨在只有局部通信而没有中央调控条件下各子系统通过局部耦合协调合作来达到整体共同目标,因此是大规模网络化局部信息交互系统控制机构的理想选择,近年来已被广泛应用于卫星姿态控制,智能交通系统,大规模电力调度等领域。协同一致性(Cooperitive Consensus)问题是网络化多智能体系统(Networked Multi-agent Systems,NMAS)分布式控制研究中最基本的问题,是解决其他分布式控制问题的基础。如何保证复杂多智能体系统在尽可能弱的通讯拓扑条件下实现高稳定度、高精度、高速度收敛是研究多智能体系统一致性问题的重点和难点。本文紧紧围绕:a)带有复杂不确定性因素的非线性多智能体系统的自适应有限时间一致性控制问题;b)多智能体系统的给定有限时间一致性控制问题,展开研究,主要完成了以下创新性工作:有限时间一致性控制(1)针对单向通讯拓扑条件下二阶非线性多智能体系统有限时间一致性控制问题,通过构造一个新的Laplacian矩阵并通过严格的理论分析证明得出该矩阵的一个重要性质,将该矩阵引入合适的Lyapunov候选函数中,提出了适用于单向通讯网络下非线性多智能体系统的分布式自适应有限时间一致性控制算法。该方法解决了现有有限时间控制理论几乎无法对其进行严格解析分析的理论难题。在此基础上,将上述有限时间稳定性分析方法进一步推广应用到该系统的编队-合围控制问题中,解决了单向拓扑条件下二阶非线性多智能体系统的连续有限时间编队-合围问题。(2)针对同时具有未知时变控制增益,不可线性参数化分解的不确定非线性因素以及未知不可测执行器故障的二阶复杂非线性多智能体系统,研究了其在单向通讯拓扑下的有限时间一致性控制问题,结合加幂积分技术,核心函数技术以及自适应方法,设计出了基于分数阶状态反馈以及分数阶自适应律的容错有限时间一致性控制算法,通过引入虚拟参数估计误差以及局部滑膜误差的概念,巧妙构造了Lyapunov候选函数,给出了系统有限时间稳定性的严格理论分析与证明,从而解决了对存在未知时变控制增益以及未知不可测执行器故障条件下复杂多智能体系统有限时间一致性控制的难题。在此基础上,进一步将该方法推广到了高阶非仿射多智能系统,结合反衍控制设计思想,归纳法思想,高阶加幂积分技术,核心函数技术以及自适应方法,建立了有限时间一致性协议,解决了高阶非仿射多智能体系统的有限时间一致性控制问题。(3)针对带有非匹配不确定性的高阶纯反馈(Pure-feedback)多智能体系统,通过引入基于虚拟分数阶滤波的分数阶-动态面设计概念,设计出了基于分数阶状态反馈的有限时间神经网络自适应一致性追踪控制算法,并通过应用Lyapunov稳定性分析方法、反衍控制设计方法以及打包函数(packaged function)技术,给出了系统稳定的严格理论分析与证明,解决了此类带有非匹配不确定性的高阶复杂系统的有限时间一致性追踪控制问题,并进一步将此稳定性分析方法拓展应用到了该类系统的有限时间合围控制中。给定有限时间控制(4)提出了一种全新的有限时间控制理论与分析方法,不同于传统的依赖符号函数或分数阶状态反馈的有限时间控制,该方法中所得的有限收敛时间不依赖于任何初始条件和其他设计参数,因此可以提前统一给定。基于该有限时间控制理论思想,针对一阶线性多智能体系统,分别在无向拓扑和有向拓扑条件下建立了在提前统一给定有限时间内实现协同一致的结果,解决了现有有限时间控制理论无法提前给定统一的有限时间表达式的难题,更进一步解决了现有有限时间控制理论不连续或不平滑的难题。更进一步,将此有限时间控制理论拓展应用到高阶多智能体系统领导-跟随一致性控制问题中,解决了高阶系统有限时间控制中存在的控制器设计复杂以及分析过程复杂导致的难以应用于实际,并且难以提前统一给定有限收敛时间的难题。
[Abstract]:Large-scale network system has complex dynamic characteristics, and its control problem is very important and challenging. Traditional centralized control method is very difficult to meet the requirement of coordination, stability and high-efficiency operation of complex large-scale system. different from the traditional centralized control method, the distributed control method has the advantages of less information amount, good cooperation, high flexibility, strong expandability and the like, It is the ideal choice of the large-scale networked local information interaction system control mechanism, which has been widely used in attitude control of satellite in recent years. Intelligent transportation system, large-scale power dispatching and other fields. Cooperative Consistency is the most basic problem in distributed control research of Networked Multi-agent Systems (NMAS), which is the basis for solving other distributed control problems. How to guarantee the high stability, high precision and high speed convergence of complex multi-agent system under the condition of weak communication topology is the key point and difficulty of studying the consistency problem of multi-agent system. This paper focuses on: a) adaptive finite time consistency control for nonlinear multi-agent systems with complex uncertainties; b) a given finite time consistency control problem for multi-agent systems, which primarily completes the following innovative work: The finite time consistency control (1) is based on the finite time consistency control problem of the second order nonlinear multi-agent system under unidirectional communication topology. By constructing a new Laplacian matrix and proving the important property of the matrix through strict theoretical analysis, In this paper, we introduce a distributed adaptive finite time consistency control algorithm for nonlinear multi-agent systems under unidirectional communication networks by introducing the matrix into appropriate Lyapunov candidate functions. The method solves the theoretical problem that the existing finite time control theory can hardly analyze and analyze the existing finite time control theory. On the basis of this, the finite time stability analysis method is further extended to the formation-joint control problem of the system, and the continuous finite time formation of the second-order nonlinear multi-agent system under one-way topology is solved. (2) Aiming at the uncertain nonlinear factors with unknown time-varying control gain and non-linear parametric decomposition and the second-order complex nonlinear multi-agent system with unknown non-measurable actuator failure, the problem of limited time consistency control in one-way communication topology is studied. Combined with power integration technique, kernel function technique and adaptive method, a finite time consistency control algorithm based on fractional order state feedback and fractional order adaptive law is designed. By introducing the concept of virtual parameter estimation error and local synovium error, The Lyapunov candidate function is ingeniously constructed, and the strict theoretical analysis and proof of the finite time stability of the system are given, thus solving the problem that the finite time consistency control of the complex multi-agent system under the condition of unknown time-varying control gain and unknown non-measurable actuator is solved. On the basis of this, the method is extended to the high-order non-affine multi-intelligence system, combined with the anti-perfunctory control design idea, the inductive method, the high-order plus power integration technique, the kernel function technology and the self-adapting method, and a finite time consistency protocol is established. and solves the problem of limited time consistency control of a high-order non-affine multi-agent system. (3) Aiming at a high order pure feedback multi-agent system with non-matching uncertainty, a fractional order-dynamic face design concept based on the virtual fractional order filtering is introduced, A finite time neural network adaptive consistency tracking control algorithm based on fractional order state feedback is designed, and the strict theoretical analysis and proof of system stability are given by applying Lyapunov stability analysis method, anti-perfunctory control design method and packaged function technique. The limited time consistency tracking control problem of the high-order complex system with non-matching uncertainty is solved, and the stability analysis method is further extended to the limited time combination control of the system. a new finite time control theory and analysis method is proposed for a given finite time control (4), which is different from the finite time control of the traditional dependency symbol function or fractional order state feedback, The finite convergence time obtained in the method does not depend on any initial conditions and other design parameters and can therefore be given uniformly in advance. based on the finite time control theory idea, aiming at one-order linear multi-agent system, solves the difficult problem that the existing finite time control theory cannot advance a uniform finite time expression in advance, and further solves the difficult problem that the existing finite time control theory is not continuous or smooth. further, the limited time control theory is extended to the leading-follow consistency control problem of the high-order multi-agent system, the complexity of the controller design existing in the finite time control of the high-order system and the complexity of the analysis process are solved, and it is difficult to unify the problem of a given finite convergence time in advance.
【学位授予单位】：重庆大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP18;TP273
本文编号：2290709
[Abstract]:Large-scale network system has complex dynamic characteristics, and its control problem is very important and challenging. Traditional centralized control method is very difficult to meet the requirement of coordination, stability and high-efficiency operation of complex large-scale system. different from the traditional centralized control method, the distributed control method has the advantages of less information amount, good cooperation, high flexibility, strong expandability and the like, It is the ideal choice of the large-scale networked local information interaction system control mechanism, which has been widely used in attitude control of satellite in recent years. Intelligent transportation system, large-scale power dispatching and other fields. Cooperative Consistency is the most basic problem in distributed control research of Networked Multi-agent Systems (NMAS), which is the basis for solving other distributed control problems. How to guarantee the high stability, high precision and high speed convergence of complex multi-agent system under the condition of weak communication topology is the key point and difficulty of studying the consistency problem of multi-agent system. This paper focuses on: a) adaptive finite time consistency control for nonlinear multi-agent systems with complex uncertainties; b) a given finite time consistency control problem for multi-agent systems, which primarily completes the following innovative work: The finite time consistency control (1) is based on the finite time consistency control problem of the second order nonlinear multi-agent system under unidirectional communication topology. By constructing a new Laplacian matrix and proving the important property of the matrix through strict theoretical analysis, In this paper, we introduce a distributed adaptive finite time consistency control algorithm for nonlinear multi-agent systems under unidirectional communication networks by introducing the matrix into appropriate Lyapunov candidate functions. The method solves the theoretical problem that the existing finite time control theory can hardly analyze and analyze the existing finite time control theory. On the basis of this, the finite time stability analysis method is further extended to the formation-joint control problem of the system, and the continuous finite time formation of the second-order nonlinear multi-agent system under one-way topology is solved. (2) Aiming at the uncertain nonlinear factors with unknown time-varying control gain and non-linear parametric decomposition and the second-order complex nonlinear multi-agent system with unknown non-measurable actuator failure, the problem of limited time consistency control in one-way communication topology is studied. Combined with power integration technique, kernel function technique and adaptive method, a finite time consistency control algorithm based on fractional order state feedback and fractional order adaptive law is designed. By introducing the concept of virtual parameter estimation error and local synovium error, The Lyapunov candidate function is ingeniously constructed, and the strict theoretical analysis and proof of the finite time stability of the system are given, thus solving the problem that the finite time consistency control of the complex multi-agent system under the condition of unknown time-varying control gain and unknown non-measurable actuator is solved. On the basis of this, the method is extended to the high-order non-affine multi-intelligence system, combined with the anti-perfunctory control design idea, the inductive method, the high-order plus power integration technique, the kernel function technology and the self-adapting method, and a finite time consistency protocol is established. and solves the problem of limited time consistency control of a high-order non-affine multi-agent system. (3) Aiming at a high order pure feedback multi-agent system with non-matching uncertainty, a fractional order-dynamic face design concept based on the virtual fractional order filtering is introduced, A finite time neural network adaptive consistency tracking control algorithm based on fractional order state feedback is designed, and the strict theoretical analysis and proof of system stability are given by applying Lyapunov stability analysis method, anti-perfunctory control design method and packaged function technique. The limited time consistency tracking control problem of the high-order complex system with non-matching uncertainty is solved, and the stability analysis method is further extended to the limited time combination control of the system. a new finite time control theory and analysis method is proposed for a given finite time control (4), which is different from the finite time control of the traditional dependency symbol function or fractional order state feedback, The finite convergence time obtained in the method does not depend on any initial conditions and other design parameters and can therefore be given uniformly in advance. based on the finite time control theory idea, aiming at one-order linear multi-agent system, solves the difficult problem that the existing finite time control theory cannot advance a uniform finite time expression in advance, and further solves the difficult problem that the existing finite time control theory is not continuous or smooth. further, the limited time control theory is extended to the leading-follow consistency control problem of the high-order multi-agent system, the complexity of the controller design existing in the finite time control of the high-order system and the complexity of the analysis process are solved, and it is difficult to unify the problem of a given finite convergence time in advance.
【学位授予单位】：重庆大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP18;TP273
【参考文献】
相关博士学位论文 前3条
1 龙晓军;多智能体系统的有限时间一致性跟踪[D];大连海事大学;2015年
2 王寅秋;非线性多智能体系统一致性分布式控制[D];北京理工大学;2015年
3 佘莹莹;多智能体系统一致性若干问题的研究[D];华中科技大学;2010年
相关硕士学位论文 前1条
1 肖秋云;多智能体系统有限时间一致性若干问题研究[D];江南大学;2015年
，本文编号：2290709
本文链接：https://www.wllwen.com/shoufeilunwen/xxkjbs/2290709.html
