Model Predictive Control leveraging Ensemble Kalman forecasting for optimal power take-off in wave energy conversion systems

Although solar and wind energy are the dominant sources of renewable energy production today, attention is increasing towards other solutions, such as marine Wave Energy Conversion (WEC). High installation costs and the relatively low efficiency of current power take-off mechanisms currently prevent WEC from being competitive with solar and wind energy production. One potential way to overcome this is by implementing an active control strategy which optimizes WEC power takeoff in real time, thereby extracting the maximum energy possible from each individual wave. Accomplishing this requires an estimation problem to be solved, since oncoming sea waves are not known a priori-they come from multiple directions, and generally arrive in sets. In the present work, a parallel Ensemble Kalman forecasting algorithm leveraging a pseudospectral wave model for time propagation of the ensemble members is developed to estimate and forecast the wavefield based on data assimilated from a Doppler wave radar system. This forecast of the future wavefield is then exploited in a linear Model Predictive Control (MPC) setting for the online optimization of power take-off of a one-body point-absorber WEC system subject to motion and machinery constraints. Simulations over a realistic sea state exhibit a trade-off between control efficiency and forecast accuracy, with power losses in the prediction-based setting not exceeding 16% of the power take-off obtained using MPC under the ideal assumption of complete knowledge of the oncoming wavefield.