Neuro-Adaptive Safe Consensus Tracking Control for Pure-Feedback Nonaffine Multiagent Systems
针对具有未知非仿射动态和外部干扰的多智能体系统，提出一种结合神经网络、动态面控制和障碍李雅普诺夫函数的分布式控制方案，确保所有智能体安全跟踪领导者路径且信号全局一致有界。
This study addresses the safe consensus tracking issue for a specific category of multiagent systems (MASs) featuring a static directed communication graph. Each follower agent is subject to external disturbances and governed by unknown pure-feedback nonaffine dynamics. To facilitate the back-stepping approach in nonaffine systems, the mean value theorem (MVT) is employed. Additionally, dynamic surface control (DSC) is implemented to mitigate the intricacies typically encountered in back-stepping frameworks. For the approximation of the unknown nonlinearities, radial basis function neural networks (NNs) are utilized. Integrating these methodologies with principles from graph theory and barrier Lyapunov functions (BLFs), we propose a tailored neuro-adaptive distributed control scheme. The objective of this scheme is to ensure that followers can accurately track the leader’s path while maintaining the globally uniformly bounded (GUB) property of all system signals within the closed loop. Comparative simulation results demonstrate the effectiveness and superiority of the proposed control method.