Fault Reconstruction Control Design for Underwater Vehicle Propeller Based on ILO-BISMC

This paper addresses single-degree-of-freedom propulsion faults in underwater vehicles caused by actuator bias errors, and proposes a fault reconstruction method that combines a learning observer with backstepping integral sliding mode control (ILO-BISMC). First, the system's nonlinear characteristics and external disturbances are analyzed, resulting in the development of a dynamic model that incorporates actuator bias faults. A sliding mode surface is then designed to address position errors, generating simulated reference velocities and speed loop inputs for control. An iterative learning observer is introduced to accurately estimate the fault data of high-frequency, time-varying bias faults in the single-degree-of-freedom system. Using this fault data, input compensation is applied to the velocity loop, ensuring passive fault-tolerant control under bias faults. The stability of the cascaded system is demonstrated using the Lyapunov stability theorem and linear matrix inequalities. Finally, pose tracking control simulations confirm the effectiveness of the proposed fault reconstruction method and validate the superiority and robustness of the control strategy.