A Neural Network-Based Power Control Method for Direct-Drive Wave Energy Converters in Irregular Waves

In this study, the maximum power extraction condition for a direct-drive wave energy converter in irregular waves in the time domain was proven and described using the variational method. Thus, a real-time optimal power control law was proposed, which contained a noncausal part. To determine this law, a classical controller requires information about the future wave excitation force. The prediction of a wave excitation force is either too costly or less accurate. This study presents a novel optimal power control method based on a back propagation (BP) neural network without wave prediction. The network was used to learn the input-output mapping relationships of the noncausal part, and it used the history data of the wave excitation force as the input. The setting and training of the network models are discussed in detail. The history data of the wave excitation force were identified using a Kalman filter. The simulation and experiment demonstrated that the proposed method was valid, effective, and superior to some existing methods.