Evaluating Machine Learning and Deep Learning models for predicting Wind Turbine power output from environmental...

Area of Science:

• Renewable Energy Systems
• Computational Intelligence
• Environmental Engineering

Background:

• Accurate wind turbine power output prediction is crucial for grid integration and energy management.
• Machine learning (ML) and deep learning (DL) offer promising approaches for enhancing forecasting accuracy.

Purpose of the Study:

• To conduct a comprehensive comparative analysis of various ML and DL models for wind turbine power output prediction.
• To evaluate model performance using key metrics such as R-squared, MAE, and RMSE.
• To identify the most effective algorithms for forecasting wind energy generation.

Main Methods:

• Evaluated a diverse set of ML models including Linear Regression, Support Vector Regressor, Random Forest, Extra Trees, Adaptive Boosting, Categorical Boosting, Extreme Gradient Boosting, and Light Gradient Boosting Machine.
• Assessed deep learning models such as Artificial Neural Network (ANN), Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), and Convolutional Neural Network (CNN).
• Utilized a dataset of 40,000 observations and applied preprocessing techniques like feature scaling and parameter tuning.

Main Results:

• Extra Trees (ET) demonstrated the highest performance among ML models, achieving an R-squared of 0.7231 and RMSE of 0.1512.
• Artificial Neural Network (ANN) showed the best performance among DL models, with an R-squared of 0.7248 and RMSE of 0.1516.
• DL models, especially ANN, exhibited slightly superior performance, indicating better capability in modeling non-linear dependencies.

Conclusions:

• Deep learning models, particularly ANN, offer enhanced predictive accuracy for wind turbine power output compared to top-performing ML models.
• The study highlights the potential of advanced computational methods for optimizing renewable energy systems.
• Preprocessing techniques significantly contribute to improving model performance and forecasting accuracy.