Flat filtering model-free control of floating offshore wind turbine above rated wind speed with radial basis function neural network

A Floating Offshore Wind Turbine (FOWT) is a highly complex nonlinear system subjected to stochastic disturbances (wind, waves, etc.). It is the technology used to harvest wind energy in deep waters and could potentially provide a substantial portion of the global electricity demand. Providing robust control for FOWTs operations is a challenge due to the complexities in modeling their dynamic behavior. Thus, this article introduces a flat filtering approach with a Radial Basis Function Neural Network (RBFNN) to a model-free adaptive control of the collective pitch angle of a FOWT operating beyond the rated wind speed (Region 3). A high-fidelity nonlinear model of a semi-submersible FOWT is used to validate the controller and compare it to the basic Gain Scheduling Proportional-Integral (GSPI) controller with Anti-Windup, and another model-free adaptive controller using Generalized Proportional Integral Control (GPIC). The performance of the proposed controller surpasses both and gives a more robust performance under high wind and waves conditions.