Bayesian adaptive framework for wind power forecasting
Nordic Machine Intelligence · 08.04.2026 · 6 (2) · 10-17
DOI: 10.5617/nmi.12911 ↗Kurz gefasst
Fachartikel in der Zeitschrift Nordic Machine Intelligence, online erschienen am 08.04.2026. Verfasst von Maria Luisa Daza Torres und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The integration of wind power into electricity grids poses significant challenges due to its inherent variability and the non-stationary nature of wind energy fleets. This paper proposes a Bayesian adaptive framework for forecasting aggregate wind power generation across Norway’s four bidding zones. Our approach employs a physics-informed nonlinear regression model where key power curve parameters are treated as time-evolving states. Using sequential Bayesian inference within a moving data window, the model continuously updates parameter estimates as new observations become available, enabling automatic adaptation to non-stationarities. Results show robust performance across heterogeneous zones, with the sequential updating mechanism successfully capturing evolving fleet dynamics and maintaining forecast reliability as new wind parks become operational. This adaptive Bayesian approach offers transmission system operators a practical tool for improved day-ahead forecasting that balances physical interpretability with statistical flexibility.
Zitieren
APA
Torres, M. L. D., Lopez, J. C. M., & Suaste, V. (2026). Bayesian adaptive framework for wind power forecasting. Nordic Machine Intelligence, 6(2), 10-17. https://doi.org/10.5617/nmi.12911
Vancouver
Torres MLD, Lopez JCM, Suaste V. Bayesian adaptive framework for wind power forecasting. Nordic Machine Intelligence. 2026;6(2):10-17. doi:10.5617/nmi.12911
BibTeX
@article{Torres2026,
author = {Torres, Maria Luisa Daza and Lopez, Jose Cricelio Montesinos and Suaste, Veronica},
title = {Bayesian adaptive framework for wind power forecasting},
journal = {Nordic Machine Intelligence},
year = {2026},
volume = {6},
number = {2},
pages = {10--17},
doi = {10.5617/nmi.12911}
}
EUVIMED