Implementation of a Model Predictive Controller (MPC) for Floating Offshore Wind Turbines (FOWT)

In recent times, renewable resources have been increasingly exploited, with wind energy being second only to hydro and solar power. Over the last decade, thanks to technological advances, the average cost of energy has come down considerably, making this sector increasingly competitive. Although onshore wind and fixed offshore are more widespread due to their greater development and fewer technological challenges, floating offshore is also starting to progress, also thanks to the possibility of exploiting areas with good wind resources but with a deep seabed and to the possibility of reaching large sizes without visual pollution or environmental impact problems. With the development of offshore floating wind turbines (OFWT), it therefore becomes necessary to adapt control strategies, starting from conventional controllers used in the non-floating counterpart, or developing new types of control, such as those based on PID logic, or unconventional ones. The aim of this thesis is the realization of a controller that falls into the latter category that is based on MPC (Model Predictive Control) logic. Controllers of this type exploit the knowledge of the mathematical model of the controlled system to find the optimal values of the control actions with the aim of maximizing its performances. After a brief introduction on the current situation of wind energy and the description of the main components of a wind turbine, the case study adopted for the implementation of the controller will be described. After that, the mathematical model of the latter will be obtained, based on a pre-existing model, and validated through it. In a second step, the control strategies for wind turbines will be analysed and a linear MPC controller will be studied from a theoretical point of view. Finally, exploiting the knowledge of the system’s model and of the theoretical foundations above mentioned, an MPC controller will be implemented, based on a set of LTI (Linear Time Invariant) systems, each of which is obtained by linearizing the starting non-linear system in different operating conditions. It will be exploited for the realization of the various controller modules, together with optimization algorithms for the search of the control actions and algorithms for the estimation and forecasting of some unknown quantities for the control loop closure. Once implemented, the controller will be tested by comparing its performance with other widely used conventional controllers, in order to draw some conclusions on its effectiveness and trying to understand how its unconventional logic can be exploited for further developments.