Relative Motion Prediction for Integrated Floating Offshore Wind Turbine Installation Scheme Based on Data-Driven

With the development of offshore wind power toward deeper seas, innovative installation methods for floating offshore wind turbines (FOWTs) are urgently needed. A novel installation approach involves a specially designed catamaran that carries multiple preassemblies (tower, hub, and blades are mated onshore) to the installation site. The approach separates the construction of foundations and preassembled upper structures. Carrying multiple preassemblies onboard shortens the transportation time, and only one lifting operation shortens the offshore mating time. Due to the complexity of the offshore environment, irregular relative motions occur between the preassemblies and the floating foundation. Larger relative distances between the mating points (lifted preassembly bottom center point and floating foundation top center point) would increase the outcrossing rate during the mating operation. Artificial neural networks are applied to predict the maximum and minimum relative motions between the two mating points during the implementation of mating operations. The installation scheme is simulated in various sea states using Sima, generating datasets for analysis. Statistical motion data (mean value and low-order central moments) for the past period are used as inputs to the neural network to obtain the maximum and minimum relative motion for the coming period. Simulation results show that the prediction model constructed by the back propagation neural network can predict maximum and minimum values of relative motions with acceptable accuracy. The proposed algorithm enables the prediction of potential hazards during the mating operation of the conceptual installation method, facilitating informed decision-making by the crews.