Development of Dynamic Model of Wave Energy Converters Using Gaussian Process Regression

Abstract

　Accurate dynamic models of wave energy converters are essential for the design of these devices. To create an accurate model, it is necessary to consider not only the free surface effects of water but also the mechanical friction forces and cogging forces of the power takeoff system. Nevertheless, these are hard to calculate theoretically. Consequently, this study addresses the development of a nonparametric dynamic model that can consider complicated forces, such as mechanical friction forces and cogging forces of a power take-off system. A regression model using a Gaussian regression process was developed to estimate the heaving displacement and velocity of a wave energy converter based on the data measured in tank tests. This regression model predicted the generated output power more accurately than a numerical simulation model based on the system identification. Additionally, it was shown that the accuracy of the regression model does not decrease significantly as the observed noise in the training data increases. An accurate dynamic model of wave energy converters based on measured data can be used for a wide range of applications, including the development of new control methods, evaluation of power production, and estimation of the maximum displacement of wave energy converters.