A genetic algorithm-based ensemble framework for wind speed forecasting.

Area of Science:

• Renewable Energy Systems
• Meteorological Forecasting
• Computational Intelligence

Background:

• Wind energy is a vital clean resource, but its variability necessitates accurate forecasting.
• Existing wind speed prediction models struggle with meteorological influences.
• Reliable forecasting is key to managing wind energy's intermittency.

Purpose of the Study:

• To develop and evaluate a genetic algorithm (GA)-based ensemble framework for enhanced wind speed forecasting.
• To systematically compare the performance of 14 diverse forecasting models.
• To assess the framework's effectiveness across multiple Brazilian cities.

Main Methods:

• A GA-based ensemble approach was proposed, combining various forecasting models.
• Fourteen models, including linear, neural network, hybrid, and ensemble types, were evaluated.
• Minute-by-minute wind speed data from five Brazilian cities were utilized for model validation.

Main Results:

• The GA-based ensemble framework demonstrated superior performance with low Mean Squared Error (MSE) and Mean Absolute Error (MAE) values.
• High R-squared ($R^2$) values (0.7139–0.8723) indicated robust predictive capabilities.
• Statistical validation (Friedman test, p < 0.001) confirmed significant model performance differences and rank stability.

Conclusions:

• The proposed GA-based ensemble framework offers a significant advancement in wind speed forecasting accuracy.
• The framework exhibits high scalability and computational efficiency, making it suitable for practical applications.
• Improved wind speed prediction enhances the integration and reliability of wind energy systems.