Resilient Control in Multi-Hydrofoil Crafts: Tackling Actuator Faults and False Data Injection for Attitude Consensus

This paper addresses the attitude consistency problem in multi-hydrofoil crafts, considering actuator faults, false data injection, stochastic ocean wave disturbances, and topological propagation effects. We propose a game-theoretic fault-tolerant control (FTC) framework integrating a composite controller with a specific performance index. In this framework, both local and neighboring node information is considered. The desired controller, along with non-ideal anomalies and disturbances, is treated as participants, transforming the FTC problem into a multi-player non-cooperative game. Moreover, a single critic neural network (NN) is utilized to alleviate the challenges associated with solving a partial differential equation, thereby facilitating the derivation of the ideal control law. Compared to traditional local information-based control methods, the proposed approach incorporates neighborhood information. It integrates faults, disturbances, and propagation effects within a unified framework for optimal control, enhancing FTC effectiveness. We conduct comparative experiments using a four-craft scenario on the dSPACE platform. The results demonstrate that the proposed method reduces the mean absolute deviation (MAD) of local tracking error by at least 28.00% and the standard deviation (STD) by at least 7.76% compared to the other two methods. Note to Practitioners—This work stems from the challenge of maintaining attitude consensus in a hydrofoil fleet operating under real-world disturbances and actuator failures, considering the propagation effects in interconnected systems. The game-theoretic framework models FTC as a competitive interaction between the controller and anomalies in a unified architecture, eliminating the need for separate fault diagnosis and improving efficiency. Additionally, NNs replace complex mathematical solutions, enabling engineers to deploy control laws without specialized knowledge of partial differential equations. Hardware-in-the-loop experiments validate the feasibility and provide a paradigm of resilience control for other interconnected control systems for pertinent practitioners.