Online Optimal Control of Wave Energy Converters via Adaptive Dynamic Programming

The control objective of wave energy converters (WECs) is to maximize energy conversion from sea waves and guarantee their safe operation. This can be expressed as a constrained optimal control problem subject to a disturbance input (the incoming wave excitation) for energy maximization. A novel energy maximization control strategy is proposed based on the idea of approximate dynamic programming (ADP), where a critic neural network (NN) is used to approximate the time-dependant optimal cost value (due to the finite-horizon cost function), whose inputs are the current system states and the time-to-go. A recently proposed adaptation based on the parameter estimation error is used to online update the weight of critic NN, where the estimation error convergence can be proved. Hence, the network output, e.g. the costate, is used to compute the optimal feedback control. The proposed WEC control strategy does not need the non-causal information of wave prediction, which makes its implementation more economically viable without significantly reducing energy output. The efficacy of the proposed WEC control approach is demonstrated using numerical simulations.