Intelligent forecasting of motion responses of offshore floating wind turbines based on artificial intelligence
Abstract
Accurate predictions of the motion response of floating wind turbine platforms under wave action are crucial for their design optimization and performance evaluation. This article proposes an intelligent forecasting method for motion response based on neural network models, which integrates convolutional layer modules, Bi-LSTM, and multi-head self-attention mechanisms to analyze the pitch and heave data of a four-column floating wind turbine platform under both regular and irregular wave conditions. The article first discusses the construction of a numerical tank and validates the effectiveness of the wave generation, propagation, and dissipation methods through numerical simulations. By forecasting the motion response data under three different operating conditions of the floating wind turbine platform with varying lead times, the efficiency and stability of the model's predictions are validated, demonstrating the accuracy and high precision of the forecasting results. This provides a fast and cost-effective forecasting solution for the real-time motion response of offshore floating wind turbine platforms, possessing both theoretical significance and important engineering value.