FFRLS-MPC: An adaptive model predictive control method for robust ship course keeping

Accurate course control is essential for safe and efficient maritime navigation. Conventional controllers often degrade under time-varying vessel dynamics and persistent environmental disturbances. We propose an adaptive course-control strategy that couples online system identification with model predictive control (MPC). The resulting FFRLS-MPC controller uses recursive least squares with a forgetting factor (FFRLS). It identifies key manoeuvring-model parameters in real time. The updated model is supplied to MPC for state prediction and constrained rudder optimisation. This loop compensates for model-plant mismatch and unmodelled disturbances. We validated the method in simulations under calm water and wind-wave-current disturbance scenarios. Performance was compared with PID and robust adaptive MPC baselines. FFRLS-MPC produced faster responses, lower steady-state errors and more accurate dynamic course tracking. The identification module captured disturbance effects as adaptive parameter variation. These results indicate a practical route to robust, high-precision ship motion control.