Distributionally Robust Bayesian Optimization for Resilient Microgrid Design Under Nonstationary Climate Uncertainty
International Journal of Energy Research · 2026-01 · 2026 (1)
DOI: 10.1155/er/6880319 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Energy Research, online erschienen am 11.11.2026. Verfasst von Bernard Atta Adjei und 4 weiteren Autor:innen. Die Arbeit nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The increasing reliance on renewable energy in microgrids introduces deep uncertainty due to nonstationary climate dynamics. Stochastic programing (SP) approaches usually assume stationary distributions and therefore cannot guarantee reliable performance under distributional shifts. In this paper, we propose a distributionally robust Bayesian optimization (DRBO) framework that jointly addresses black‐box system design and the ambiguity in environmental distributions. Microgrid planning is formulated as a min–max problem over a Wasserstein ambiguity set that captures the uncertainty in renewable generation and load, and it is solved with a Gaussian process (GP)‐based Bayesian optimization (BO) scheme. The proposed approach not only optimizes the worst‐case expected cost but also examines the coordinated design and operational behavior across multiple energy resources. Numerical experiments on a synthetic, climatologically calibrated testbed, in which a structured adversary searches for renewable droughts, load surges, and gradual drift, show that the Wasserstein radius acts as an explicit control of conservatism. Five‐fold cross‐validation selects a small radius, which lowers the held‐out tail cost (CVaR 0.95 ) by 3% relative to no robustness and reduces its variability across folds about fivefold. On 50 shifted test scenarios, the DRBO design has a 4.6% higher expected cost than SP but a 6.8% lower CVaR 0.95 and no loss of load, whereas chance‐constrained and deterministic designs are cheaper on average but have heavier cost tails. At the same capital cost as the SP design, DRBO finds a design with a lower expected and tail cost, and the dispatch analysis shows that DRBO invests in an electrolyzer–tank–fuel‐cell chain that converts surplus wind into a growing hydrogen reserve.
Zitieren
APA
Adjei, B. A., Appiah, S. T., Salia, J., Dzikunu, C. N., & Ankamah, J. D. (2026). Distributionally Robust Bayesian Optimization for Resilient Microgrid Design Under Nonstationary Climate Uncertainty. International Journal of Energy Research, 2026(1). https://doi.org/10.1155/er/6880319
Vancouver
Adjei BA, Appiah ST, Salia J, Dzikunu CN, Ankamah JD. Distributionally Robust Bayesian Optimization for Resilient Microgrid Design Under Nonstationary Climate Uncertainty. International Journal of Energy Research. 2026;2026(1). doi:10.1155/er/6880319
BibTeX
@article{Adjei2026,
author = {Adjei, Bernard Atta and Appiah, Sampson Takyi and Salia, Jessica and Dzikunu, Charles Nunya and Ankamah, Johnson De-Graft},
title = {Distributionally Robust Bayesian Optimization for Resilient Microgrid Design Under Nonstationary Climate Uncertainty},
journal = {International Journal of Energy Research},
year = {2026},
volume = {2026},
number = {1},
doi = {10.1155/er/6880319}
}
EUVIMED