WIND TURBINE POWER PREDICTION USING MACHINE LEARNING MODELS: A CASE STUDY WITH REAL FARM DATA

The power generated from wind turbines is of critical importance as one of the fundamental components of sustainable and renewable energy systems. However, the complex and nonlinear nature of wind flow and the influence of interconnected factors make turbine power estimation significantly difficult. This study aims to evaluate the performance of different forecasting models using real-time data obtained from wind turbines and to determine the most effective model for wind power generation. The analyses are performed based on performance metrics that measure the agreement between the predicted and actual values. The study results reveal that the Decision Tree Regressor model provides the highest accuracy with 0.998 R² value and low error rates (RMSE: 0.151, MAE: 0.036) and that tree-based models are more effective in wind power estimation. These models, trained using large datasets, offer significant potential in terms of increasing power grid stability and ensuring the optimization of wind farms. The study shows that advanced methods used in turbine power estimation are an effective tool for optimizing renewable energy use by contributing to sustainable energy targets.