Answers are grounded in this paper and cite their sources.
Wind energy forecast validation is crucial for the reliable integration of renewable energy into the power grid. This study evaluates and compares the short-term wind power forecasts from two institutions: Administración Nacional de Usinas y Transmisiones Elóctricas (UTE) and Administración del Mercado Elóctrico (ADME). In collaboration with UdelaR-Fing, UTE has developed three models and also used forecasts provided by “Meteologica,” while ADME utilizes “Meteoblue” forecasts adjusted by a proprietary model. The model employs neural networks to refine wind speed predictions and then uses a sigmoid-like exponential function to convert these predictions into power forecasts. This compares the forecast accuracy using Mean Absolute Error (MAE). Preliminary results indicate that for the year 2023 Meteologica’s forecasts have lower mean absolute error than ADME, suggesting a higher forecasting ability. It also shows that ADME has a more centered model while Meteorological tends to underestimate it. Future work will explore power percentiles over a more extended period and the impact of forecast initialization times on error metrics.
No takes yet. Share an insight, caveat, or question.
Ungerovich et al. (2024) studied this question.
Synapse has enriched one closely related paper. Consider it for comparative context: