多智能体系统一致性是近十年来力学、物理、交通、控制等领域的研究热点。本项目研究仅基于相对输出测量信息的分布式固定时间观测器与自适应一致性协议设计问题及其在多飞行器分布式姿态一致性追踪与蔡氏混沌电路同步问题中的应用。针对带不确定非线性的多智能体系统，明确所需观测量的物理意义，提出基于相对输出测量信息的分布式固定时间观测器设计的理论框架，给出仅与系统动态相关的观测器存在性的充分必要条件，揭示其与传输零点之间的关系；提出分布式固定时间降阶观测器的设计方法，讨论最低阶次分布式固定时间观测器的维数与系统维数及相对阶之间的关系；基于分布式观测器设计完全分布式自适应一致性协议，克服外部信号或未知非线性的影响，并应用于多飞行器分布式姿态一致性追踪及蔡氏混沌电路的同步问题中。本项目的研究将建立不依赖于通讯而仅需测量的一致性问题的新方法，从根源上消除网络攻击隐患，为实际重要领域的应用奠定理论基础。

The consensus of multi-agent systems has been a hot topic of various fields such as Mechanics, Physics, Transportation and Control in recent decade. This project studies distributed fixed-time observer and the corresponding adaptive consensus protocol design based on only relative output measurement among neighboring agents, and the applications on consensus tracking of attitude angles for multiple aircrafts as well as the chaos synchronization of Chua’s circuits. For the multi-agent systems with uncertain nonlinearities, we first determine which variable should be observed, and then present the structure of designing distributed fixed-time observers with only relative output measurement. Furthermore, the existence of the designed distributed fixed-time observer is studied to give sufficient and necessary conditions based on system dynamics, and the relationship between transmission zero condition and the proposed conditions are revealed. Then, the reduced-order distributed fixed-time observers are designed, and the relationship among the dimension of the minimal-order distributed fixed-time observer, the dimension of system dynamics, as well as the relative degree is presented. Based on distributed fixed-time observers, fully distributed adaptive protocols are designed to realize consensus for multi-agent systems with external disturbances or unknown nonlinearities. The proposed adaptive protocols are further applied into consensus tracking of attitude angles for multiple aircrafts as well as the chaos synchronization of Chua’s circuits. The study of this project would present novel framework on consensus protocol design based on only relative output measurement, without using any communication information or eigenvalue information of Laplacian matrix. In this sense, the proposed protocols can radically eliminate the potential network attacks, and provide theoretic supports for practical applications.