Privacy‐Preserving and Interpretable Stability Prediction in Smart Grids Using Explainable Federated Feature Selection
International Journal of Energy Research · 2026-01 · 2026 (1)
DOI: 10.1155/er/1942591 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Energy Research, online erschienen am 20.08.2026. Verfasst von Muhammad Adnan Khan und 7 weiteren Autor:innen. Die Arbeit nennt 54 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The increasing integration of renewable energy sources and distributed smart devices within modern power systems has intensified the challenges of real‐time load balancing and grid stability management. Traditional centralized prediction models struggle to adapt to dynamic load fluctuations and introduce significant privacy concerns by requiring access to sensitive, user‐level energy data. Moreover, the black‐box nature of many machine learning (ML) systems undermines stakeholder trust and limits the ability of grid operators to obtain actionable insights. To address these limitations, this research presents a privacy‐preserving, interpretable, and feature‐efficient framework for grid stability prediction, leveraging federated learning (FL), feature selection, and explainable artificial intelligence (XAI). The novelty of this study lies in integrating federated feature selection, global model training, and local interpretable model‐agnostic explanation (LIME)‐based explanation into a unified workflow for privacy‐aware and interpretable grid stability prediction. The proposed model decentralizes the learning process, enabling distributed grid nodes to collaboratively train energy management models without sharing raw data. This approach preserves data confidentiality while maintaining predictive performance. A federated feature‐selection mechanism is incorporated to identify the most relevant features across participating clients, thereby increasing efficiency and generalizability without jeopardizing privacy. Additionally, LIME is integrated to generate local explanations, thereby improving transparency and enabling stakeholders to understand the decision logic in real time. Experimental evaluations on benchmark grid datasets confirm that the feature‐selected federated‐XAI model not only achieves high predictive accuracy (99.49%) and a minimal miss rate (0.51%) but also delivers robust explainability and privacy safeguards. This approach demonstrates a scalable, secure, and trustworthy solution for smart grid stability prediction and decision support in the next generation. The predicted stability status can further support downstream operational decisions such as load balancing and preventive grid control.
Zitieren
APA
Khan, M. A., Anis, G., Naz, N. S., Saleem, M., Mazhar, T., Farooq, M. S., Lee, S., & Hamam, H. (2026). Privacy‐Preserving and Interpretable Stability Prediction in Smart Grids Using Explainable Federated Feature Selection. International Journal of Energy Research, 2026(1). https://doi.org/10.1155/er/1942591
Vancouver
Khan MA, Anis G, Naz NS, Saleem M, Mazhar T, Farooq MS, et al. Privacy‐Preserving and Interpretable Stability Prediction in Smart Grids Using Explainable Federated Feature Selection. International Journal of Energy Research. 2026;2026(1). doi:10.1155/er/1942591
BibTeX
@article{Khan2026,
author = {Khan, Muhammad Adnan and Anis, Gulfaraz and Naz, Naila Samar and Saleem, Muhammad and Mazhar, Tehseen and Farooq, Muhammad Sajid and Lee, Sang-Woong and Hamam, Habib},
title = {Privacy‐Preserving and Interpretable Stability Prediction in Smart Grids Using Explainable Federated Feature Selection},
journal = {International Journal of Energy Research},
year = {2026},
volume = {2026},
number = {1},
doi = {10.1155/er/1942591}
}
EUVIMED