GENCO meets friends: Foundation Model Mixing for a Full Dynamic All-Electric Society
Working paper, code and data of the submission to the GridFM Challenge with GENCO (Porto 2026): the grid foundation model GENCO and classical state estimation (WLS) find the available capacity of every line and transformer of the UK Power Networks grid Ninfield for the next day, with Timestral-209M as forecast. This is a working paper, and results may still change. Files genco-meets-friends_working-paper.pdf: the working paper (also in the zip) genco-meets-friends-v1.zip, folder genco-meets-friends/: 0_paper_documentation/: working paper, data quality, data lineage, figures 1_data/: Ninfield data and the stored predictions of every path as Parquet, with sources, licences, checksums; preparation/: steps from the raw data 2_configuration/: all settings, frozen before the test 3_models/: WLS-AC, WLS-DC, GENCO, Timestral-209M call 4_results_evaluation/: one script per table and figure, checks Run: pip install -r requirements.txt, python 4_results_evaluation/check_post_numbers.py --models. Code Apache License 2.0, text and data CC BY 4.0. Contains data from UK Power Networks Open Data Portal, licensed under CC BY 4.0; modified. Weather data by Open-Meteo.com (CC BY 4.0). Research prototype, not for grid operation, provided without warranty.
Authors
- Marco Obermeier (ORCID: https://orcid.org/0009-0004-9178-4353)
Institutions
- University of Würzburg (DE)
- Weihenstephan-Triesdorf University of Applied Sciences (DE)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-26
- DOI
- https://doi.org/10.5281/zenodo.23622825
- Primary Topic
- Thermal Analysis in Power Transmission
- Type
- article
- Field-Weighted Citation Impact
- 0.00