Enhancing Short-Term Wind Speed Prediction Based on Deep Learning With Ensemble Learning Model for Small Wind Turbine Applications

The short-term wind speed prediction is critical for efficient wind energy production, particularly in small wind turbine applications. This study presents an enhanced prediction model based on deep learning integrated with an ensemble learning approach. The system employs a combination of machine learning algorithms, including XGBoost, MLP, and LSTM, to predict wind speed and assess the feasibility of wind energy generation. The model leverages nine key meteorological features, such as air temperature, wind speed, and turbulence intensity, to provide highly accurate predictions. The backend of the system is built with Flask, which integrates a trained XGBoost model for real-time predictions. The system allows users to upload CSV data for analysis, register and log in to access the model’s functionalities, and view the feasibility of wind energy production. Additionally, the model's accuracy is evaluated, demonstrating exceptional performance in wind speed prediction. This tool provides significant support for decision-making in the renewable energy sector, especially for small-scale wind turbine installations, ensuring efficient use of resources and optimizing energy output. By leveraging advanced deep learning techniques combined with ensemble learning models, this research enhances the reliability and robustness of wind speed forecasting for sustainable energy production.

Keywords:

Short-term wind speed prediction, deep learning, ensemble learning, XGBoost, MLP, LSTM, wind energy feasibility, small wind turbines, renewable energy, Flask.