Answers are grounded in this paper and cite their sources.
The control of Floating Offshore Wind Turbines (FOWTs) in Region III is challenging due to complex aerodynamic, hydrodynamic, and structural interactions. This paper presents a fully data-driven, model-free Deep Reinforcement Learning (DRL) controller based on the Trust Region Policy Optimization (TRPO) algorithm to regulate the collective blade pitch of a 5 MW semi-submersible FOWT. The controller was trained in high-fidelity simulations and experimentally validated in a wave basin using a Software-In-the-Loop (SIL) approach. Results show improved generator speed regulation and platform stability compared to a baseline Gain-Scheduling Proportional–Integral (GSPI) controller. However, performance degradation with generator speed overshoots was observed under extreme wind conditions. This study highlights the potential of DRL for FOWT control and identifies future directions to enhance robustness in harsh environments.
No takes yet. Share an insight, caveat, or question.
Didier et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: