Optimal control techniques can oversee the power take off (PTO) operation of wave energy conversion systems to ensure that the overall power output is maximized, but optimization in real-time poses difficulties given the wave variability and the underlying constraints of the system. This study examines several different model predictive control (MPC) approaches that utilize a model of the hydrodynamics of the wave energy converter as well as the dynamics of a hydraulic PTO system. The impact of accounting for various dynamics in the optimization and the role of constraints on the wave energy converter performance is explored for irregular wave conditions.


As global energy demands and climate concerns continue to grow, the need for a broader range of renewable energy options is becoming increasingly clear. While wave energy converters (WECs) have been researched for decades, much work has focused on the design of such systems. Many different structures have emerged over the years including point absorbers, attenuators, oscillating water columns and reservoirs (Aderinto and Li, 2018). Point absorber WEC systems have a mass that is moved up and down by the waves and this motion is used in some manner to drive a generator and produce electricity. The maximum energy will be captured if the frequency of the WEC system matches the dominant frequency of the incoming wave (i.e., resonance) and this is typically achieved by a PTO system that can effectively add or remove damping and thereby affect the device's frequency. A wide range of PTO systems exist including turbines, hydraulic systems, and linear actuators (Têtu, 2017).

Controlling the PTO system remains challenging due to the growing complexity of WEC systems and the variable nature of the incoming waves. Early strategies often manipulated the PTO system by leveraging linear models and velocity tracking, complex conjugate approaches or impedance matching control. Studies such as these of Hals et al., 2011a and Garcia-Violini et al., 2020 have compared a variety of these techniques. While these methods have merits, they often encounter challenges with operation over a wide frequency range and at times suffered from a high computational burden. Those with feedforward components also needed wave excitation estimation and as such, were more prone to performance problems due to wave prediction errors (García-Violini, et al., 2020).

This content is only available via PDF.
You can access this article if you purchase or spend a download.