Assessing ANN Architectures for Wind Turbine Power Prediction: A Comparative Study

With the increasing importance of renewable energy sources, wind power has emerged as a significant contributor to the global energy mix. Accurate prediction of wind power production is essential for efficient grid integration and optimal utilization of wind resources. This study presents an investigation into the performance of various artificial neural network (ANN) models for wind power prediction. Different ANN architectures, including narrow, medium, and wide networks, as well as bilayered and trilayered structures, were explored to understand their impact on predictive capabilities. The data used for evaluation was collected from a 2.05 MW Senvion MM82 wind turbine. The performance of the ANN models was compared with Linear Regression (LR), Interactions LR, Robust LR, and Stepwise LR methods. Results revealed that the bilayered and trilayered ANN s achieved the best performance in wind power prediction. This study highlights the potential of ANN models in accurately predicting wind power, thereby facilitating efficient and reliable wind farm operations.