Model predictive control of a single-buoy wave energy converter with coupled constraints and model adaptation

Abstract

A novel optimization framework is introduced for a single-buoy wave energy converter (WEC), which integrates model predictive control (MPC) with the boundary element method (BEM). A time-series auto-regressive (AR) model is employed to predict the wave excitation force in the near future for the MPC algorithm, and its accuracy is evaluated. Incorporation of an AR wave prediction model within the MPC framework significantly enhances operational efficiency and reduces computational costs. Application of a quadratic viscous damping correction and quantification of the prediction horizon effectively mitigated model mismatch between the linearized state-space idealization utilized within the optimizer and the actual controlled WEC. Mitigation of model mismatch significantly enhances prediction accuracy and broadens the applicability of the MPC. A sensitivity analysis is performed, and the coupled effects between constraints are examined to determine an optimal performance benchmark. The imposition of constraints on power take-off (PTO) force variation within the MPC framework ensures system resilience, indicating the practical feasibility of the approach. Simulations based on conditions typical of Zhaitang Island, China, suggest that employing an MPC algorithm for the WEC can increase maximum energy capture efficiency by 20% compared to optimal passive control. The MPC is shown to adjust the device behavior in response to varying sea conditions and to optimize performance for each sea state.