• An EKF is implemented to perform online parameter estimation of the nominal MPC model in the presence of measurement noise. • The integral action is explicitly embedded within the MPC framework, enabling systematic selection of suitable integral gains and the implementation of anti-windup strategies. • A novel anti-windup mechanism is introduced to prevent excessive integral accumulation while maintaining continuous integral behavior. This study presents an extended Kalman filter (EKF)-based adaptive integral model predictive control (MPC) strategy for the depth regulation of an autonomous underwater vehicle (AUV) equipped with a moving mass control (MMC) system. The EKF is employed to enable real-time estimation of uncertain system parameters under measurement noise, thereby ensuring continuous online model adaptation. To compensate for steady-state errors caused by the angle of attack, an integral action is explicitly embedded within the MPC framework, which allows for systematic tuning of the integral gain as well as the incorporation of anti-windup strategies. Furthermore, a novel adaptive anti-windup mechanism is proposed to suppress excessive integral accumulation while preserving continuous integral dynamics, thereby improving control smoothness and robustness in the presence of actuator saturation and hysteresis. The closed-loop stability of the proposed controller is rigorously analyzed and theoretically guaranteed. Comprehensive simulation and experimental results demonstrate the effectiveness of the proposed approach.
Chai et al. (Wed,) studied this question.