Auto-tuning of Non-Linear Model Predictive Controllers for a Remote Operated Vehicle

In this work, an auto-tuning algorithm has been proposed to automatically obtain the weight adjustment parameter of one variable in a Nonlinear model predictive controller (NMPC). The NMPC is part of the double loop controller implemented in conjunction with an integral action.Furthermore, disturbances corresponding to model noise, sensor noise and second order waves effects have been applied to the remotely operated vehicle (ROV) nonlinear model, including during controller tuning. In adittion, an Unscented Kalman Filter (UKF) has been implemented for model state estimation and filtering of the disturbances that affect the model output.Finally, simulations have been carried out in which by observing the system output, it can be concluded that the proposed autotuning algorithm generates an adequate tuning of the control parameter.