基于切换方法的网络化系统分布式事件触发控制研究

It is of important significance in theory and application to study distributed event-triggered control of networked systems due to its advantages such as sufficient resource utilization, design flexibility, strong fault tolerance and so on. However, it is a challenging task to analyze and synthesize such systems because of the simultaneous existence of the nonlinear coupling among the subsystems, event-triggering mechanism and network topology changes. This project conducts a research on distributed event-triggered control of networked systems with nonlinear coupling: (1) A hybrid time-delay system model is proposed to characterize coupling continuous dynamics, triggered discrete events and variable topologies. (2) Under a fixed topology, a new scheme of distributed controller and event-triggering condition co-design and stability analysis is established by combining Lyapunov-like approaches, a small-gain theorem and an algebraic graph theory. (3) Under a varying topology, an approach to co-design of distributed controllers and event-triggered conditions is proposed by making use of multiple Lyapunov functions and state space partition methods in switched systems. (4) The validness and practical application of theoretical results obtained are verified by building an experiment platform for distributed event-triggered control of networked multi-wheeled robots. This project reveals the intrinsic relationship among event-triggered conditions, controller design and variable topologies, and provides an effective method to co-design distributed controllers and event-triggered conditions of networked systems and the foundation for its pratical application.

由于网络化系统分布式事件触发控制具有资源利用率高、设计灵活和容错性强等优点，其研究具有重要的理论意义和应用价值。然而，非线性耦合、事件触发机制和网络拓扑结构变化等并存，给此类系统的分析和综合带来极大挑战。项目针对具有耦合的网络化系统分布式事件触发控制展开研究：（1）建立混杂时滞系统模型，表征耦合的连续动态、事件触发的离散动态和变拓扑等特性；（2）融合类Lyapunov函数、小增益定理和图论，给出固定拓扑结构下稳定性分析以及分布式控制器和触发条件联合设计方案；（3）采用切换系统多Lyapunov函数和状态空间划分法，实现变拓扑结构下分布式控制器和事件触发条件联合设计；（4）构建网络化多轮式机器人分布式事件触发控制实验平台，检验提出方法的可应用性。项目揭示事件触发条件、控制器设计和变拓扑结构之间的内在联系，为网络化系统分布式控制器和事件触发条件联合设计提供有效方法，为其实际应用奠定基础。

随着网络化系统规模的不断增大，集中式控制全局信息难于获取，且容错性差难于维护。而分布式控制策略利用局部资源信息采取决策行为，通过有效的协调机制实现系统全局性能指标，具有较强的灵活性和容错性等优点，受到普遍关注。网络化系统普遍存在信息传输时滞和数据丢失等现象，严重破坏系统控制性能。项目研究了事件触发条件与控制器设计内在联系，构造新的Lyapunov函数，给出固定拓扑结构和切换拓扑结构下基于事件触发机制的分布式包含控制和优化控制协议设计方案。研究了网络化系统传输时滞、数据包丢失等因素下控制综合问题，给出观测器和控制器的存在条件。进一步地，研究了有向通讯拓扑结构下多智能体系统分布式优化协议设计，克服信息传递不对称性带来的困难。最后，基于深度视觉等信息搭建多机器人协作控制实验平台，实现多轮式机器人变队形控制和双臂机器人协作抓取，为军事和民用推广奠定基础。围绕上述研究工作，在本领域重要期刊和会议发表和录用学术论文52篇，其中SCI检索期刊高水平论文29篇，顶级期刊Automatica论文1篇，IEEE Transactions期刊论文11篇，SCI他引近100次。项目负责人以第一完成人身份获得第七届中国人工智能学会吴文俊人工智能自然科学二等奖，中国控制会议关肇直优秀论文提名奖。申请发明专利3项，授权软件著作权3项。与美国和新加坡学者建立长期合作关系，培养博士毕业生1名，硕士毕业生15名，在读博士研究生5名，硕士研究生15名，多名研究生获得国家级、省级和校级奖励。