| description abstract | The dynamic performance prediction of floating offshore wind turbines (FOWTs) is a challenging task, as the existing theories might not be fully reliable for FOWTs due to the high nonlinearities and coupling effects. The artificial intelligent (AI) method gives a promising solution for this issue, and Chen and Hu proposed a novel AI-based method, named SADA (software-in-the-loop combined artificial intelligence method for dynamic response analysis of FOWTs), to overcome these challenges. This paper addresses a further and in-depth investigation of the key technologies of the key disciplinary parameters (KDPs) in the SADA method to obtain a novel and accurate analysis method for dynamic responses prediction of FOWTs. First, the categorization of KDPs is introduced, which can be divided into three categories: environmental KDPs, disciplinary KDPs, and specific KDPs. Second, two factors, the number of KDPs and boundary adjustment of KDPs, are investigated through the reinforcement learning algorithm within the SADA method. Basin experimental data of a spar-type FOWT is used for AI training. The results show that more proper KDPs set in the SADA method can lead to higher accuracy for the prediction of FOWTs. Besides, reasonable boundary conditions will also contribute to the convergence of the algorithms efficiently. Finally, the instruction on how to better choose KDPs and how to set and adjust their boundary conditions is given in the conclusion. The application of KDPs in the SADA method not only provides a deeper understanding of the dynamic response of the entire FOWTs system but also provides a promising solution to overcome the challenges of validation. | |
