Neural network-based dynamic target enclosing control for uncertain nonlinear multi-agent systems over signed networks

Weihao Li; Jiangfeng Yue; Mengji Shi; Boxian Lin; Kaiyu Qin

doi:10.1016/j.neunet.2024.107057

Neural network-based dynamic target enclosing control for uncertain nonlinear multi-agent systems over signed networks

Neural Netw. 2024 Dec 20:184:107057. doi: 10.1016/j.neunet.2024.107057. Online ahead of print.

Authors

Weihao Li¹, Jiangfeng Yue², Mengji Shi³, Boxian Lin⁴, Kaiyu Qin⁵

Affiliations

¹ School of Aeronautics and Astronautics, University of Electronic Science and Technology of China, Chengdu, 611731, Sichuan, China; Aircraft Swarm Intelligent Sensing and Cooperative Control Key Laboratory of Sichuan Province, Chengdu, 611731, Sichuan, China. Electronic address: [email protected].
² School of Aeronautics and Astronautics, University of Electronic Science and Technology of China, Chengdu, 611731, Sichuan, China; Aircraft Swarm Intelligent Sensing and Cooperative Control Key Laboratory of Sichuan Province, Chengdu, 611731, Sichuan, China. Electronic address: [email protected].
³ School of Aeronautics and Astronautics, University of Electronic Science and Technology of China, Chengdu, 611731, Sichuan, China; Aircraft Swarm Intelligent Sensing and Cooperative Control Key Laboratory of Sichuan Province, Chengdu, 611731, Sichuan, China. Electronic address: [email protected].
⁴ School of Aeronautics and Astronautics, University of Electronic Science and Technology of China, Chengdu, 611731, Sichuan, China; Aircraft Swarm Intelligent Sensing and Cooperative Control Key Laboratory of Sichuan Province, Chengdu, 611731, Sichuan, China. Electronic address: [email protected].
⁵ School of Aeronautics and Astronautics, University of Electronic Science and Technology of China, Chengdu, 611731, Sichuan, China; Aircraft Swarm Intelligent Sensing and Cooperative Control Key Laboratory of Sichuan Province, Chengdu, 611731, Sichuan, China. Electronic address: [email protected].

PMID: 39721102
DOI: 10.1016/j.neunet.2024.107057

Abstract

Neural networks have significant advantages in the estimation of uncertainty dynamics, which can afford highly accurate prediction outcomes and enhance control robustness. With this in mind, this study presents a neural network-based method to investigate the uncertain target enclosing control problem for multi-agent systems over signed networks. Firstly, a nominal target enclosing controller is constructed by adding the target information component into the classical bipartite consensus error, in which the multi-agent system can be grouped to enclose the target from opposite sides. Secondly, the uncertain dynamics of the target and matched/unmatched disturbances of agents are estimated to generate the feedforward control components by adopting the neural network approximation. Therefore, high-cost sensors are unnecessary for applications that require obtaining high-order information about a target, such as velocity and acceleration, while still ensuring accurate target-enclosing control. Additionally, the proposed target enclosing controller exhibits improved robustness in the presence of both matched and unmatched disturbances. To further demonstrate its effectiveness, numerical simulations are conducted.

Keywords: Matched/unmatched disturbances; Multi-agent systems; Neural networks; Signed networks; Target enclosing.