多智能体系统一致性控制若干问题研究

发布时间：2018-11-25 12:30

【摘要】：随着计算机、信息、自动化等学科的发展,多智能系统的协调控制在实际工程中展示出广阔的应用前景。一致性问题是多智能体协调控制的基础问题,现已成为系统与控制科学领域的研究热点之一。本文以图论、矩阵分析和Lyapunov稳定性理论为工具,围绕多智能体系统一致性控制中存在的若干问题展开研究。本文的主要内容如下:1.研究了具有通讯时延的二阶非线性多智能体在时间控制切换拓扑下的一致性问题。通过引入状态变换,得到降阶的一致性误差系统,于是多智能体系统的一致性问题转化为误差系统的稳定性问题。最后,构造Lyapunov-Krasovskii泛函推导出一致性协议的设计方法并得到拓扑切换信号的最小平均驻留时间。2.研究了二阶非线性多智能体系统在状态控制切换拓扑下的一致性问题。文中引入状态变换,并证明了二阶多智能体系统可由一个一阶多智能体系统等价地分析。首先,在固定拓扑情况下,得到一致性协议的设计方法。然后,设计状态控制的拓扑切换规则,并构造共同Lyapunov函数,证明在所设计的一致性协议和拓扑切换规则作用下,二阶非线性多智能体在切换拓扑联合包含生成树的情况下可达到一致。3.研究了高阶非线性多智能体系统在复杂环境下的一致性问题。一方面,研究了智能体在执行器受到Backlash-like磁滞影响并且控制方向未知情况下的一致性问题。文中利用Backlash-like磁滞的性质,将磁滞输出分解为线性输入和有界扰动的叠加。接着,采用新型的Nussbaum函数,解决了多智能体未知控制方向的问题。最后,利用反步法,并结合神经网络,推导出分布式自适应神经网络一致性协议。另一方面,研究了具有状态时延的高阶非线性多智能体系统一致性问题。基于反步法,利用神经网络补偿不确定非线性项,提出自适应一致性协议。同时,通过估计神经网络权重矩阵的上界,减少了自适应参数的数量,从而降低了计算负荷。最终,设计Lyapunov-Krasovskii泛函,证明了所提出一致性协议的有效性。4.研究了异质多智能体系统的分组一致性,并考虑通讯拓扑中存在代表竞争关系的负权重边。首先,考虑一类半异质的多智能体系统,即只允许不同分组中的智能体的动力学模型是异质的。利用线性系统的小增益定理,设计了分布式状态反馈的分组一致性协议。然后,考虑一类异质多智能体系统的分组一致性,这时所有智能体的动力学模型都允许是不同的,结合输出调节理论和小增益定理,分别提出基于状态反馈和动态输出反馈两种分组一致性协议。最终揭示组内耦合关系和组间耦合关系对多智能体系统分组一致性的影响。
[Abstract]:With the development of computer, information, automation and other disciplines, the coordinated control of multi-intelligent system has shown a broad application prospect in practical engineering. The consistency problem is the basic problem of multi-agent coordinated control, and has become one of the hotspots in the field of system and control science. Using graph theory, matrix analysis and Lyapunov stability theory as tools, this paper focuses on several problems in consistency control of multi-agent systems. The main contents of this paper are as follows: 1. The consistency problem of a second order nonlinear multi-agent with communication delay under time-controlled switching topology is studied. By introducing the state transformation, a reduced order consistency error system is obtained, so the consistency problem of the multi-agent system is transformed into the stability problem of the error system. Finally, the Lyapunov-Krasovskii functional is constructed to derive the design method of the consistency protocol and the minimum mean dwell time of the topology switching signal is obtained. 2. 2. In this paper, the consistency problem of second order nonlinear multi-agent systems under state control switching topology is studied. In this paper, the state transformation is introduced, and it is proved that the second-order multi-agent system can be analyzed equally by a first-order multi-agent system. Firstly, in the case of fixed topology, the design method of conformance protocol is obtained. Then, the topology switching rules of state control are designed, and the common Lyapunov function is constructed. The second order nonlinear multi-agent is consistent in the case of switching topology joint inclusion spanning tree. 3. 3. The consistency problem of high order nonlinear multi-agent systems in complex environments is studied. On the one hand, the consistency problem of the agent is studied when the actuator is affected by the Backlash-like hysteresis and the control direction is unknown. By using the properties of Backlash-like hysteresis, the hysteresis output is decomposed into the superposition of linear input and bounded disturbance. Then, the new Nussbaum function is used to solve the problem of unknown control direction of multi-agent. Finally, a distributed adaptive neural network consistency protocol is derived by using the backstepping method and the neural network. On the other hand, the consistency problem of high order nonlinear multi-agent systems with state delay is studied. Based on the backstepping method, an adaptive consistency protocol is proposed by using neural networks to compensate for uncertain nonlinear terms. At the same time, by estimating the upper bound of the neural network weight matrix, the number of adaptive parameters is reduced and the computational load is reduced. Finally, the Lyapunov-Krasovskii functional is designed to prove the effectiveness of the proposed conformance protocol. 4. 4. The packet consistency of heterogeneous multi-agent systems is studied, and the negative weight edges representing competition in communication topology are considered. Firstly, we consider a class of semi-heterogeneous multi-agent systems, that is, the dynamic model of agents in different groups is heterogeneous. Based on the small gain theorem of linear systems, a distributed state feedback packet consistency protocol is designed. Then, considering the grouping consistency of a class of heterogeneous multi-agent systems, the dynamic models of all agents are allowed to be different, combined with the output regulation theory and the small gain theorem. Two packet consistency protocols based on state feedback and dynamic output feedback are proposed respectively. Finally, the effects of intra-group coupling and inter-group coupling on packet consistency in multi-agent systems are revealed.
【学位授予单位】：广东工业大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TP18;TP13

【参考文献】