Short-term wind speed forecasting in Uruguay using computational intelligence.

Area of Science:

• Renewable Energy Systems
• Artificial Intelligence in Meteorology
• Wind Resource Assessment

Background:

• Accurate short-term wind speed forecasting is crucial for integrating wind power into national grids.
• Existing forecasting methods may have high computational costs, necessitating efficient alternatives.
• Uruguay aims to enhance its renewable energy infrastructure, requiring reliable wind energy predictions.

Purpose of the Study:

• To develop and validate an artificial neural network (ANN) technique for short-term wind speed forecasting.
• To assess the ANN's predictive accuracy at various heights using real-world data.
• To provide a cost-effective and accurate forecasting tool for wind power integration in Uruguay.

Main Methods:

• Collected one year of hourly wind speed data from anemometers at multiple heights (101.8, 81.8, 25.7, 10.0 m).
• Applied and optimized different artificial neural network (ANN) configurations for each measurement height.
• Conducted quantitative analysis and statistical evaluation to select the best-performing ANN models.

Main Results:

• The artificial neural network (ANN) technique demonstrated high accuracy across all tested anemometer heights.
• The optimized ANN models effectively predicted real wind speed data.
• The computational costs of the ANN method were found to be lower than numerical modeling.

Conclusions:

• The proposed ANN method is a powerful and accurate tool for short-term wind speed forecasting in Uruguay.
• This approach offers a scientifically significant contribution to reliable large-scale wind power forecasting and grid integration.
• The method can assist the Administración Nacional de Usinas y Transmissiones Eléctricas in managing the national energy supply effectively.