Comparative Analysis of Long Short-Term Memory and Transformer for Wind Speed Prediction: Performance, Challenges, and Optimization

This study compares the performance of Long Short-Term Memory and Transformer models in predicting wind speed, optimized using the hybrid PSO-SA algorithm. Accurate wind speed prediction is critical for efficient wind farm management, reducing carbon emissions and reliance on fossil fuels. Hourly wind speed, air pressure, relative humidity, and air temperature data from the Alta Wind Energy Center (2020–2022) were used. Key hyperparameters for both models were optimized, and ensemble methods like Random Forest, XGBoost, CatBoost, and Stacking were applied. Random Forest emerged as the most accurate model, demonstrating superior performance in integrating LSTM and Transformer outputs. This research emphasizes the strengths and limitations of each approach and underscores the role of advanced machine learning techniques in enhancing the efficiency and reliability of wind power generation.