多智能体系统基于结构分解的分析与控制

The study of multi-agent systems has important theoretical significance and application value in many fields such as engineering and biology. The deep understanding of the internal structure of the multi-agent system is a key step to perfect the theoretical framework of the multi-agent system. Hence, the project is devoted to the study of multi-agent distributed systems in the framework of structural decomposition. Firstly, the project will focus on exploring the effective structure decomposition approaches and the formal characteristics of the system structure decomposition, as well as the system performance based on the internal structure decomposition.Then the project will study the identification and construction of the local topology which can destroy the system performance. Based on this, the project will find out the internal relationship between the decomposition of the system internal structure and the design of the distributed protocol and gain. Finally, as a key target of docking the previous research, a system structure decomposition method will be established with respect to the decentralized stabilization problem, as well as a protocol design method will be developed based on structural decomposition. Then the system stabilization methodology will be built based on the structural decomposition. By taking advantage of the characteristics of the internal structure of multi-agent systems, the project will propose new methods for the analysis and control of multi-agent systems and discover new rules of dynamic evolution within the system.

多智能体系统的研究在工程、生物等诸多领域具有重要的理论意义和应用价值。对多智能体系统内部结构的深入认识是完善多智能体系统理论框架的关键一步，因此，项目致力于在结构分解的框架下对多智能体系统进行研究。首先，研究多智能体系统结构分解的途径和形式特点，以及基于内部结构分解的系统性能。进而，研究破坏系统性能的局部拓扑结构的甄别与构造；然后，在此基础上，发现系统内部结构分解与分布式协议和增益设计之间的内在关系；最后，作为前述研究的一个重点对接目标，将针对分布式镇定问题建立相关的系统结构分解形式，以及基于结构分解的协议设计方法，并进而发展系统结构分解基础上的镇定方法。项目将从多智能体系统的内部结构特征出发，提出适用于多智能体系统分析和控制的新方法，发现系统内部动态演化的新规律。

群体智能是人工智能的一个重要领域，作为群体智能研究中的一个主要研究对象，多智能体系统及其动态演化规律的理解，受到了当今学者的普遍关注。本项目研究了多智能体系统的分布式镇定问题和分布式能控性问题的图理论特征刻画，获得了基于图拓扑特征的判定系统性能的判据和代数条件，以及多智能体网络化系统的一致性问题和网络化控制系统的最优控制问题。取得了一批国际水准的研究成果，部分成果具原创意义，发表在国际顶级期刊上。特别地，提出并建立了多智能体系统分布式能控性和可镇定性基于图拓扑特征的判据，和分析求解的一般步骤，并刻画了判据求解及系统实现的复杂性。此外，给出了系列线性及部分非线性多智能体系统实现分布式一致性的充要条件，成果涉及快速收敛、时变时滞、有向拓扑、切换拓扑等相应内容。共发表论文50余篇，包括33篇SCI国际期刊论文和17篇中文核心期刊论文和中国卓越期刊论文，包括领域顶级期刊Automatica上的长文和知名期刊IEEE Transactions on Cybernetics 、SCIENCE CHINA: Information Sciences以及International Journal of Robust and Nonlinear Control上发表的文章。