Artificial Intelligence-Based Comparative Analysis of Wind Energy Forecasting Models

Accurate wind energy predicting is essential for optimizing power grid stability, energy trading, and renewable energy integration. This study explores the performance of eight machine learning models-Linear Regression, Random Forest, Support Vector Machine (SVM), Artificial Neural Network (ANN), XGBoost, LightGBM, CatBoost, and Long Short-Term Memory (LSTM)-to predict wind power generation. The models were evaluated based on key performance metrics, including Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> score. Experimental results indicate that ensemble learning methods, particularly XGBoost, LightGBM, CatBoost, and Random Forest, achieved near-perfect predictions with an <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> score of 1.0000, significantly outperforming other models. ANN also demonstrated strong predictive capabilities with an <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> score of 0.9853, whereas traditional regression and SVM struggled with higher errors. LSTM, expected to excel in time-series forecasting, failed with an <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> score of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{- 0. 9 1 2 6}$</tex>, showing poor generalization. The findings highlight the superiority of ensemble models for wind energy forecasting, providing a robust solution for improving grid efficiency and renewable energy utilization.