Excitation forces estimation for non-linear wave energy converters: A neural network approach

Investigating optimal control algorithms is a continuing concern within the Wave Energy field. A considerable amount of literature has been published on optimal control architectures applied to Wave Energy Converter (WEC) devices. However, most of them requires the knowledge of the wave excitation forces acting on the WEC body. In practice such forces are unknown and an estimate must be used. In this work a methodology to estimate the wave excitation forces of a non-linear WEC along with the achievable accuracy, is discussed. A feedforward Neural Network (NN) is applied to address the estimation problem. Such a method aims to map the WEC dynamics to the wave excitation forces by training the network through a supervised learning algorithm. The most challenging aspects of these techniques are the ability of the network to estimate data not considered in the training process and their accuracy in presence of model uncertanities. Numerical simulations under different irregular sea conditions demonstrate accurate estimation results of the NN approach as well as a small sensitivity to changes in the plant parameters relative to the case study presented.