Software-in-the-Loop Combined Reinforcement Learning Method for Dynamic Response Analysis of FOWTs

Floating offshore wind turbines (FOWTs) still face many challenges on how to better predict the dynamic responses. Artificial intelligence (AI) brings a new solution overcome these with intelligent strategies. A AI technology-based method, named SADA, is proposed in this paper for prediction of responses FOWTs. Firstly, methodology SADA introduced selection Key Disciplinary Parameters (KDPs). The module was built coupled aero-hydro-servo-elastic in-house program DARwind and policy decision provided by machine learning algorithms deep deterministic gradient (DDPG). Secondly, set basin experimental results Hywind Spar-type FOWT were employed train module. weights KDPs DDPG algorithms' actor network changes their values according training feedback 6DOF motions platform through comparing simulation that data. Many other cannot be measured experiment could predicted higher accuracy . Finally, case study method conducted demonstrated mean platform's can AI-based accuracy, example maximum error surge motion reduced 21%. This takes advantage numerical-experimental which promising overcoming handicap impeding direct use traditional technology FOWTs design.