Adaptive Fault-Tolerant Formation Control of Unmanned Underwater Vehicles Based on Predictive Compensation and Credibility Assessment

Résumé fourni par la source

Abstract This paper addresses the formation control problem of unmanned underwater vehicles (UUVs) under multiple challenges including time-varying communication delays, degradation of state information reliability, and complex ocean disturbances. A fault-tolerant control strategy based on predictive compensation and credibility assessment is proposed. The framework adopts a hierarchical architecture. First, a joint compensator integrating an extended state observer (ESO) and receding-horizon prediction is developed to estimate and compensate communication delays and external disturbances online. Second, a credibility assessment module based on a deep auto-encoder (DAE) dynamically generates information weights through reconstruction errors, enabling anomaly detection and smooth fusion of predictive information. Finally, a sliding mode–PID hybrid controller is designed, which automatically switches control modes according to the vector acceleration error, ensuring strong transient robustness while achieving high-precision steady-state tracking. By constructing a composite Lyapunov function, uniform ultimate boundedness of all closed-loop signals is rigorously guaranteed. Simulation results demonstrate that the proposed method significantly outperforms conventional leader–follower and single robust control schemes in terms of tracking accuracy, stability, and fault tolerance under complex scenarios involving delays, disturbances, and intermittent information anomalies.