{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,30]],"date-time":"2026-10-30T17:36:03Z","timestamp":1793381763753,"version":"4.5.0"},"update-to":[{"DOI":"10.12688\/openreseurope.24355.1","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2026,8,6]],"date-time":"2026-08-06T00:00:00Z","timestamp":1785974400000}}],"reference-count":31,"publisher":"CERN","license":[{"start":{"date-parts":[[2026,8,6]],"date-time":"2026-08-06T00:00:00Z","timestamp":1785974400000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"European Commission","award":["101135927"],"award-info":[{"award-number":["101135927"]}],"id":[{"id":"https:\/\/ror.org\/00k4n6c32","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Open Res Europe"],"abstract":"<jats:p>Background\n                    Federated Learning (FL) enables collaborative model training across distributed clients without sharing raw data, making it attractive for privacy-sensitive energy forecasting. However, standard aggregation algorithms, most notably FedAvg and FedProx, assume stationary client data distributions, an assumption routinely violated in multi-national power grids where seasonal patterns, demand crises, and policy changes induce persistent regime shifts.\n                \n                \n                    Methods\n                    We present RAFA (Regime-Aware Federated Aggregation), a novel FL framework that explicitly detects and adapts to distribution shifts in client model updates without accessing raw client data. RAFA comprises three components. These are a Mahalanobis-distance detector operating on random-projected update vectors; reliability-weighted aggregation that exponentially discounts shifted clients while preserving exact FedAvg behaviour when no shift is detected; and always-on personalised head fine-tuning that adapts each client\u2019s final prediction layer to its local regime.\n                \n                \n                    Results\n                    We validate RAFA on a new multi-national benchmark of six years (2019\u20132024) of hourly electrical load data for eight European bidding zones sourced from the ENTSO-E Transparency Platform. RAFA achieves a test MAE of 185.4\u00a0MW, an 11.8% improvement over FedAvg (210.2\u00a0MW), with by far the largest gain for the client exhibiting the strongest seasonal contrast (France, \n                        \n\n                            \n                                \u2212\n                                30.5\n                                %\n                            \n) and consistent improvements across the remaining clients (e.g., Portugal \n                        \n\n                            \n                                \u2212\n                                7.6\n                                %\n                            \n, Poland \n                        \n\n                            \n                                \u2212\n                                7.4\n                                %\n                            \n). Ablation studies confirm both mechanisms contribute independently.\n                \n                \n                    Conclusions\n                    RAFA provides a lightweight, privacy-preserving extension of standard federated aggregation that is robust to regime shifts in energy systems. The approach generalises to any federated setting with temporally non-stationary client distributions and a separable model architecture.<\/jats:p>","DOI":"10.12688\/openreseurope.24355.2","type":"journal-article","created":{"date-parts":[[2026,8,6]],"date-time":"2026-08-06T11:39:10Z","timestamp":1786016350000},"update-policy":"https:\/\/doi.org\/10.68336\/openreseurope.updatepolicy","source":"Crossref","is-referenced-by-count":0,"title":["Regime-Aware Federated Aggregation (RAFA) for privacy-preserving energy load forecasting across heterogeneous European grid clients"],"prefix":"10.68336","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-1886-8284","authenticated-orcid":true,"given":"Mahmoud","family":"Abbasi","sequence":"first","affiliation":[{"name":"AIR Institute, Salamanca, Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alfonso","family":"Gonz\u00e1lez Briones","sequence":"additional","affiliation":[{"name":"BISITE Research Group, University of Salamanca, Salamanca, Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0005-3808-8399","authenticated-orcid":true,"given":"Alesandro","family":"G\u00f3mez Villar","sequence":"additional","affiliation":[{"name":"BISITE Research Group, University of Salamanca, Salamanca, Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Javier","family":"Curto Hern\u00e1ndez","sequence":"additional","affiliation":[{"name":"BISITE Research Group, University of Salamanca, Salamanca, Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8175-2201","authenticated-orcid":true,"given":"Javier","family":"Prieto","sequence":"additional","affiliation":[{"name":"BISITE Research Group, University of Salamanca, Salamanca, Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"13904","published-online":{"date-parts":[[2026,8,6]]},"reference":[{"key":"115082","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.20625657"},{"key":"115083","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.20626328"},{"key":"115084","doi-asserted-by":"publisher","DOI":"10.3390\/en16052283"},{"key":"115085","doi-asserted-by":"publisher","DOI":"10.1145\/3678182"},{"key":"115086","unstructured":"Arivazhagan MG, Aggarwal V, Singh AK, et al.: Federated Learning with Personalization Layers. arXiv Preprint arXiv:1912.00818. 2019."},{"key":"115087","doi-asserted-by":"crossref","unstructured":"Ben Ali M, El-Rifai O, Megdiche I, et al.: A Survey on Clustered Federated Learning: Taxonomy, Analysis and Applications. arXiv Preprint arXiv:2501.17512. 2025.","DOI":"10.1145\/3843202"},{"key":"115088","unstructured":"Blanchard P, El Mhamdi EM, Guerraoui R, et al.: Machine Learning with Adversaries: Byzantine Tolerant Gradient Descent. Advances in Neural Information Processing Systems (NeurIPS). 2017; Vol. 30."},{"key":"115089","unstructured":"Box GEP, Jenkins GM, Reinsel GC, et al.: Time Series Analysis: Forecasting and Control. Wiley; 5th ed.2015."},{"key":"115090","doi-asserted-by":"publisher","DOI":"10.1007\/s11042-021-11219-x"},{"key":"115091","unstructured":"Olivares C, C\u00e9sar C, Brockmeier AJ: A Systematic Analysis of Out-of-Distribution Detection Under Representation and Training Paradigm Shifts. arXiv Preprint arXiv:2511.11934. 2026."},{"key":"115092","doi-asserted-by":"publisher","DOI":"10.1016\/j.engappai.2025.110980"},{"key":"115093","unstructured":"ENTSO-E: ENTSO-E Transparency Platform. 2024. Reference Source (https:\/\/transparency.entsoe.eu)"},{"key":"115094","doi-asserted-by":"crossref","unstructured":"Gao H, He Z, Jian P: Detecting OOD Samples via Optimal Transport Scoring Function. Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). 2025.","DOI":"10.1109\/ICASSP49660.2025.10890502"},{"key":"115095","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijforecast.2003.09.015"},{"key":"115096","doi-asserted-by":"crossref","unstructured":"Hosseinalipour S, Li S, Inaolaji A, et al.: Synergies Between Federated Foundation Models and Smart Power Grids. arXiv Preprint arXiv:2509.16496. 2025.","DOI":"10.1109\/MELE.2025.3639354"},{"key":"115097","unstructured":"International Energy Agency: World Energy Outlook 2022. IEA;2022. Reference Source (https:\/\/www.iea.org\/reports\/world-energy-outlook-2022)"},{"key":"115098","unstructured":"Jimenez G, Daniel M, Solans D, et al.: Non-IID Data in Federated Learning: A Survey with Taxonomy, Metrics, Methods, Frameworks and Future Directions. arXiv Preprint arXiv:2411.12377. 2024."},{"key":"115099","doi-asserted-by":"publisher","DOI":"10.1090\/conm\/026\/737400"},{"key":"115100","unstructured":"Jothimurugesan E, Hsieh K, Wang J, et al.: Federated Learning Under Distributed Concept Drift. Proceedings of the 26th International Conference on Artificial Intelligence and Statistics (AISTATS). 2023; 5834\u201353."},{"key":"115101","doi-asserted-by":"publisher","DOI":"10.1007\/s10462-025-11223-9"},{"key":"115102","unstructured":"Lee K, Lee K, Lee H, et al.: A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks. Advances in Neural Information Processing Systems (NeurIPS). 2018; Vol. 31."},{"key":"115103","unstructured":"Li T, Sahu AK, Zaheer M, et al.: Federated Optimization in Heterogeneous Networks. Proceedings of Machine Learning and Systems (MLSys). 2020; 2: 429\u2013450."},{"key":"115104","unstructured":"Liang PP, Liu T, Ziyin L, et al.: Think Locally, Act Globally: Federated Learning with Local and Global Representations. arXiv Preprint arXiv:2001.01523. 2020."},{"key":"115105","unstructured":"Liu Y, Hu T, Zhang H, et al.: iTransformer: Inverted Transformers Are Effective for Time Series Forecasting. Proceedings of the 12th International Conference on Learning Representations (ICLR). 2024."},{"key":"115106","unstructured":"Mavromatis I, De Feo S, Khan A: FLAME: Adaptive and Reactive Concept Drift Mitigation for Federated Learning Deployments. Proceedings of the ACM EWSN 2024 \u2013 EMERGE Workshop. 2024."},{"key":"115107","unstructured":"McMahan B, Moore E, Ramage D, et al.: Communication-Efficient Learning of Deep Networks from Decentralized Data. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS). 2017; 1273\u201382."},{"key":"115108","doi-asserted-by":"crossref","unstructured":"Rahman R, Kumar N, Nguyen DC: Electrical Load Forecasting in Smart Grid: A Personalized Federated Learning Approach. Proceedings of the IEEE Consumer Communications & Networking Conference (CCNC). 2025a.","DOI":"10.1109\/CCNC54725.2025.10976072"},{"key":"115109","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2025.3586115"},{"key":"115110","unstructured":"Ren J, Fort S, Liu J, et al.: A Simple Fix to Mahalanobis Distance for Improving Near-OOD Detection. arXiv Preprint arXiv:2106.09022. 2021."},{"key":"115111","unstructured":"Yin D, Chen Y, Kannan R, et al.: Byzantine-Robust Distributed Learning: Towards Optimal Statistical Rates. Proceedings of the 35th International Conference on Machine Learning (ICML). 2018; 5650\u20135659."},{"key":"115112","doi-asserted-by":"publisher","DOI":"10.1145\/3760788"}],"container-title":["Open Research Europe"],"original-title":[],"link":[{"URL":"https:\/\/open-research-europe.ec.europa.eu\/articles\/6-228\/v2\/xml","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ore.eu\/ore\/article\/download\/1423\/version\/2383\/4763","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ore.eu\/ore\/article\/download\/1423\/version\/2383\/4764","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ore.eu\/ore\/article\/download\/1423\/version\/2383\/4763","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,30]],"date-time":"2026-10-30T17:12:36Z","timestamp":1793380356000},"score":1,"resource":{"primary":{"URL":"https:\/\/ore.eu\/ore\/article\/view\/1423\/version\/2383"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,6]]},"references-count":31,"URL":"https:\/\/doi.org\/10.12688\/openreseurope.24355.2","relation":{"is-version-of":[{"id-type":"doi","id":"10.12688\/openreseurope.24355.1","asserted-by":"subject"}],"has-review":[{"id-type":"doi","id":"10.21956\/openreseurope.26842.r80193","asserted-by":"object"}]},"ISSN":["2732-5121"],"issn-type":[{"value":"2732-5121","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,6]]},"assertion":[{"value":"2026-08-06","name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}},{"value":"This project has received funding from the European Union\u2019s Horizon Europe research and innovation programme under grant agreement No [101135927] [\u201cA catalyst for EuropeaN ClOUd Services in the era of data spaces, high-performance and edge computing (NOUS)\u201d].   \n\n                    The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.","name":"funding_statement","label":"Funding Statement","group":{"name":"funding_information","label":"Funding Information"}}],"article-number":"228"}}