文摘
This paper presents a predictive optimization-based model reference adaptive control (MRAC) approach for dynamic positioning (DP) of a fully actuated underwater vehicle subject to dynamic uncertainties and actuator saturation. Compared with conventional linear reference model-based approaches, this proposed MRAC controller utilizes an optimized reference model composed of the closed-loop approximate vehicle model under a nonlinear model predictive controller, in which both the state and input constraints are considered. An adaptive dynamic inversion controller is designed to track the reference trajectory in the presence of dynamic uncertainties, and a single hidden layer neural network is incorporated to compensate for the mismatch of the actual and approximate models and ensure the convergence of tracking errors. The effectiveness of the proposed DP approach is validated by comparative simulations performed with a remotely operated vehicle.