Data-driven security control for unknown nonlinear MASs with hybrid faults: A hierarchical control approach.

Abstract

In this paper, a novel hierarchical data-driven consensus control strategy is developed for unknown nonlinear multi-agent systems (MASs) subject to hybrid faults. Unlike conventional model-free adaptive control (MFAC) methods that rely on consensus errors, the proposed approach introduces a hierarchical framework that structurally decouples the control process, thereby enhancing robustness and scalability. First, a fully distributed observer is employed to estimate the leader's dynamics based solely on locally available real-time input-output measurements, without relying on any prior knowledge of the system model. Then, a distributed MFAC-based controller is designed and embedded with an online actuator fault estimation mechanism to handle unknown faults in real time. This hierarchical design enables each agent to operate independently, reduces inter-agent interference, and streamlines the adjustment of controller parameters. Moreover, theoretical analysis ensures that all estimation and tracking errors remain uniformly bounded under hybrid fault conditions. Finally, simulation studies on two numerical MAS examples and multi-manipulator platforms demonstrate the effectiveness and practical applicability of the proposed method.