Bu Xuhui, Liang Jiaqi, Hou Zhongsheng, Chi Ronghu
IEEE Trans Neural Netw Learn Syst. 2021 May;32(5):1963-1973. doi: 10.1109/TNNLS.2020.2995600. Epub 2021 May 3.
This article considers the problem of finite-time consensus for nonlinear multiagent systems (MASs), where the nonlinear dynamics are completely unknown and the output saturation exists. First, the mapping relationship between the output of each agent at the terminal time and the control input is established along the iteration domain. By using the terminal iterative learning control method, two novel distributed data-driven consensus protocols are proposed depending on the input and output saturated data of agents and its neighbors. Then, the convergence conditions independent of agents' dynamics are developed for the MASs with fixed communication topology. It is shown that the proposed data-driven protocol can guarantee the system to achieve two different finite-time consensus objectives. Meanwhile, the design is also extended to the case of switching topologies. Finally, the effectiveness of the data-driven protocol is validated by a simulation example.
本文研究了非线性多智能体系统（MASs）的有限时间一致性问题，其中非线性动力学完全未知且存在输出饱和。首先，沿着迭代域建立了每个智能体在终端时刻的输出与控制输入之间的映射关系。通过使用终端迭代学习控制方法，根据智能体及其邻居的输入和输出饱和数据，提出了两种新颖的分布式数据驱动一致性协议。然后，针对具有固定通信拓扑的MASs，推导了与智能体动力学无关的收敛条件。结果表明，所提出的数据驱动协议能够保证系统实现两种不同的有限时间一致性目标。同时，该设计也扩展到了切换拓扑的情况。最后，通过一个仿真例子验证了数据驱动协议的有效性。
