异构多智能体系统的一致性研究

发布时间：2018-06-03 22:00

本文选题：异构多智能体系统 + 一致性　；参考：《华北电力大学(北京)》2017年硕士论文

【摘要】：目前,多智能体系统在传感器结构配置,网络拥塞控制,编队控制、航空航天探索等众多领域得到广泛应用,作为其重点研究内容的分布式协调控制也成为众多学者的关注焦点。其中一个重要的问题就是一致性问题。在没有全局控制与整体通信的情况下,系统达成一致的关键在于设计局部智能体之间的相互作用方式,使得智能体个体通过获知的邻居智能体的信息反复调整自身的行为,令所有智能体最终或在有限时间内实现一致性。在实际情况里,各种限制条件的存在或者实际任务要求,系统中个体的动态可能不尽相同。因此,异构多智能体系统的研究重若丘山,而目前在该系统一致性方面的研究成果相对较少。所以,本文致力于对异构多智能体系统及伴随其他复合问题时一致性的探求。本文基于相关学者研究成果,运用稳定性理论、控制理论、数学机理等相关知识,对异构系统达成一致性所需结构及算法参数进行探讨。首先,基于离散时间模型,本文展开对异构系统达成一致需满足的条件的探讨,着重分析系统最终的平衡状态。对于异构系统在无时延和伴随有界传输时延时的情况,分别基于有向固定拓扑和切换拓扑进行了分析。通过状态变换,分析整合成的系统矩阵,将异构多智能体系统最终的一致平衡点量化。将最终的平衡状态更直观的用系统的拉普拉斯矩阵的特征向量和初始状态表示,得到的结果简洁明了。进一步,将相关分析方法应用于系统含有有界传输时延的情况,同样获得了系统实现一致性的充分条件。其次,当系统为离散时间模型,对含有时延的异构系统一致性进行了的频域分析。对系统在无时延,伴随输入时延,既有输入时延又含有传输时延的三种情况分别进行了分析。为实现研究系统的一致性,针对一阶、二阶动态个体设计相应的一致性协议。在研究的过程中,将系统在时域下的模型转换到频域模型下,主要运用代数图论、稳定性理论和频域分析方法针对有无时延和不同拓扑等情况进行了一致性分析。最后,将关注的焦点聚焦在异构系统的分组一致性相关问题上。有针对性的对一阶动态和二阶动态智能体量身打造分组一致性控制策略。利用降阶变换的方法,使得异构系统的分组一致性问题等价于同构系统的稳定性问题。当系统无时延或含有有界传输时延时,利用代数图论和稳定性理论,给出了系统在固定拓扑下实现有界一致性的拓扑条件和参数要求。当系统含有一致有界输入时延时,构造合适的李雅普诺夫函数,系统实现一致性的充分条件根据求解的线性矩阵不等式来获取。
[Abstract]:At present, multi-agent system is widely used in many fields, such as sensor configuration, network congestion control, formation control, aerospace exploration and so on. Distributed coordination control, which is the focus of its research, has also become the focus of attention of many scholars. One of the important issues is consistency. In the absence of global control and overall communication, the key to reach agreement is to design the interaction between local agents so that the agent repeatedly adjusts its behavior through the information of the neighbor agent. Make all agents eventually or in a limited time to achieve consistency. In the actual situation, the individual dynamics in the system may be different due to the existence of various constraints or the actual task requirements. Therefore, the research of heterogeneous multi-agent system is heavy, but the research results of the system consistency are relatively few. Therefore, this paper is devoted to the research of the consistency of heterogeneous multi-agent systems and other composite problems. Based on the research results of relevant scholars, this paper discusses the structure and algorithm parameters needed to achieve consistency in heterogeneous systems by using stability theory, control theory, mathematical mechanism and other relevant knowledge. Firstly, based on the discrete time model, this paper discusses the conditions that need to be met in order to reach an agreement on heterogeneous systems, focusing on the analysis of the final equilibrium state of the system. In the case of heterogeneous systems with no delay and with bounded transmission delay is analyzed based on directed fixed topology and handoff topology respectively. Through the state transformation, the integrated system matrix is analyzed, and the final uniform equilibrium point of the heterogeneous multi-agent system is quantified. The final equilibrium state is more intuitively represented by the eigenvector and the initial state of the Laplace matrix of the system, and the result is simple and clear. Furthermore, the correlation analysis method is applied to the case of the system with bounded transmission delay, and the sufficient condition for the consistency of the system is also obtained. Secondly, when the system is a discrete time model, the consistency of heterogeneous systems with time delay is analyzed in frequency domain. In this paper, three cases of system with no delay, with input delay, both input delay and transmission delay, are analyzed. In order to realize the consistency of the research system, the corresponding consistency protocols are designed for the first and second order dynamic individuals. In the process of research, the system model in time domain is converted to frequency domain model, and the consistency analysis is mainly carried out by using algebraic graph theory, stability theory and frequency domain analysis method for the existence of delay and different topologies. Finally, the focus of attention will be focused on the heterogeneous system of packet consistency related issues. A group consistency control strategy is designed for first and second order dynamic agents. By using the method of reduced order transformation, the group consistency problem of heterogeneous systems is equivalent to the stability problem of isomorphic systems. When the system has no delay or contains bounded transmission, the topological conditions and parameter requirements for the system to realize bounded consistency under fixed topology are given by using algebraic graph theory and stability theory. When the system contains uniformly bounded input, the sufficient conditions for the system to achieve consistency are obtained according to the linear matrix inequalities (LMIs), and a suitable Lyapunov function is constructed.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】