{"version":"1","article":{"id":"private-ev-charging-forecasting-secfd-fl-2026","slug":"private-ev-charging-forecasting-secfd-fl-2026","language":"en","url":"https://haberpaper.com/en/news/private-ev-charging-forecasting-secfd-fl-2026","api_url":"https://haberpaper.com/api/v1/articles/private-ev-charging-forecasting-secfd-fl-2026?lang=en","title":"Study combines federated learning and secure aggregation for EV charging forecasts","summary":"Researchers in the United States and Canada presented SecFD-FL, a method for forecasting electric-vehicle charging loads without centralizing distributed data. The study tested forecasts and gradient-inversion attacks on Palo Alto station records and EV WATTS data.","category":"energy","region":"world","topics":["ai"],"published_at":"2026-10-13T20:11:52+03:00","modified_at":"2026-10-13T20:11:52+03:00","publication_basis":"publication-record","event_period":"2026-10-09","event_period_basis":"event-or-reference-period","sources":[{"name":"arXiv, SecFD-FL: Frequency-Decoupled Federated Learning with Secure Aggregation for Charging Load Forecasting in Internet of Electric Vehicles (arXiv:2610.12909)","url":"https://arxiv.org/abs/2610.12909","checkedAt":"2026-10-13","publishedAt":"2026-10-09"}],"translations":[{"language":"tr","url":"https://haberpaper.com/tr/news/private-ev-charging-forecasting-secfd-fl-2026","api_url":"https://haberpaper.com/api/v1/articles/private-ev-charging-forecasting-secfd-fl-2026?lang=tr"},{"language":"en","url":"https://haberpaper.com/en/news/private-ev-charging-forecasting-secfd-fl-2026","api_url":"https://haberpaper.com/api/v1/articles/private-ev-charging-forecasting-secfd-fl-2026?lang=en"},{"language":"ar","url":"https://haberpaper.com/ar/news/private-ev-charging-forecasting-secfd-fl-2026","api_url":"https://haberpaper.com/api/v1/articles/private-ev-charging-forecasting-secfd-fl-2026?lang=ar"},{"language":"fr","url":"https://haberpaper.com/fr/news/private-ev-charging-forecasting-secfd-fl-2026","api_url":"https://haberpaper.com/api/v1/articles/private-ev-charging-forecasting-secfd-fl-2026?lang=fr"},{"language":"es","url":"https://haberpaper.com/es/news/private-ev-charging-forecasting-secfd-fl-2026","api_url":"https://haberpaper.com/api/v1/articles/private-ev-charging-forecasting-secfd-fl-2026?lang=es"}],"data_types":[],"data":null,"data_compilations":[],"data_downloads":[],"content":{"paragraphs":["A preprint posted on arXiv on 9 October by researchers at Tennessee Tech in the United States and the University of Ottawa in Canada introduces SecFD-FL, a method addressing both variable demand and privacy in electric-vehicle charging forecasts. It combines a dual-stream GRU that models fast fluctuations separately from longer-term trends with zero-sum masking intended to keep raw client data from being exposed to the server during aggregation. Tests used records from five municipal charging stations in Palo Alto and selected vehicle histories from EV WATTS.","Against six federated-learning baselines, the paper reports R² of 0.847 on the Palo Alto data and 0.927 on EV WATTS, and up to a 60% reduction in global RMSE compared with centralized baselines. The authors also report less data similarity under their tested gradient-inversion attack than with raw FedAvg updates, without added per-round communication. These results apply to two datasets and the authors’ threat model; the masking does not establish absolute protection against every privacy attack, and forecasting gains have not been field-validated on a live charging network. The work is a preprint that has not yet undergone peer review."],"key_points":[],"sections":[]},"text":"Researchers in the United States and Canada presented SecFD-FL, a method for forecasting electric-vehicle charging loads without centralizing distributed data. The study tested forecasts and gradient-inversion attacks on Palo Alto station records and EV WATTS data.\n\nA preprint posted on arXiv on 9 October by researchers at Tennessee Tech in the United States and the University of Ottawa in Canada introduces SecFD-FL, a method addressing both variable demand and privacy in electric-vehicle charging forecasts. It combines a dual-stream GRU that models fast fluctuations separately from longer-term trends with zero-sum masking intended to keep raw client data from being exposed to the server during aggregation. Tests used records from five municipal charging stations in Palo Alto and selected vehicle histories from EV WATTS.\n\nAgainst six federated-learning baselines, the paper reports R² of 0.847 on the Palo Alto data and 0.927 on EV WATTS, and up to a 60% reduction in global RMSE compared with centralized baselines. The authors also report less data similarity under their tested gradient-inversion attack than with raw FedAvg updates, without added per-round communication. These results apply to two datasets and the authors’ threat model; the masking does not establish absolute protection against every privacy attack, and forecasting gains have not been field-validated on a live charging network. The work is a preprint that has not yet undergone peer review.","corrections":[],"editorial_illustration":{"kind":"symbolic-editorial-illustration","documentary":false,"scope":"section","asset_id":"energy","url":"https://haberpaper.com/assets/illustrations/energy-1536.webp","width":1536,"height":1024,"alt":"Symbolic energy landscape with solar panels, wind turbines and transmission lines.","language":"en"},"section_illustration":{"kind":"symbolic-editorial-illustration","documentary":false,"scope":"section","asset_id":"energy","url":"https://haberpaper.com/assets/illustrations/energy-1536.webp","width":1536,"height":1024,"alt":"Symbolic energy landscape with solar panels, wind turbines and transmission lines.","language":"en"}}}