Energy-Efficient Explainable Graph Intelligence Framework for Cyber Physical Smart Grid Stability Prediction
International Journal of Computational Intelligence Systems · 22.08.2026 · 19 (1)
DOI: 10.1007/s44196-026-01547-x ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computational Intelligence Systems, online erschienen am 22.08.2026. Verfasst von T. Siron Anita Susan und 6 weiteren Autor:innen. Die Arbeit nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This work proposes an Energy-Efficient Graph Intelligence Framework (EEGF) to accurately predict smart grid stability in highly dynamic operating conditions, while maintaining interpretability and resilience. The growing share of renewable energy, distributed energy resources, and cyber-physical interactions have made it much more complex to keep the grid stable, efficient, and cyber adaptable. To overcome the above challenges, the proposed framework combines Physics-Informed Graph Neural Networks (PI-GNN), Graph Attention Learning (GAT), Federated Edge Intelligence (FEI), Bayesian Hyperparameter Optimization (BHO) and Explainable Artificial Intelligence (XAI) as a single energy-efficient decision-making framework. The physics-informed learning module keeps power system operational constraints, and graph attention models the nonlinear spatial relationship between the interconnected power system components. Federated learning offers a scalable option that does not require centralizing data information, and Bayesian optimization determines the optimum parameter for learning. The explainable intelligence layer offers more transparent models and more operationally confident by adding physically meaningful feature attribution through an explainable layer based on SHAP. The framework is tested with the open Electrical Grid Stability Simulated Data Set under a variety of operating situations, such as uncertainty in communication, uncertainty in loading, uncertainty in renewable generation and contingency scenarios. Experimental investigations show that the overall accuracy, precision, recall, F1-score, AUC-ROC, Matthews Correlation Coefficient, balanced accuracy and the inference latency of the proposed model are 99.18%, 99.06%, 99.11%, 99.08%, 0.9987, 0.983, 99.09% and 18.7ms respectively and are better than the Random Forest, XGBoost, LightGBM, CatBoost, GraphSAGE and Transformer-based baseline models. The framework proposed in this paper is scalable, interpretable and computation-efficient solution for next generation of intelligent smart grid monitoring, real-time stability of the grid and secure operational decision support.
Zitieren
APA
Susan, T. S. A., Marshell, M. J., Chandra, B., Pandi, M., Rajagopal, S., Babu, C. N. K., & Li, J. (2026). Energy-Efficient Explainable Graph Intelligence Framework for Cyber Physical Smart Grid Stability Prediction. International Journal of Computational Intelligence Systems, 19(1). https://doi.org/10.1007/s44196-026-01547-x
Vancouver
Susan TSA, Marshell MJ, Chandra B, Pandi M, Rajagopal S, Babu CNK, et al. Energy-Efficient Explainable Graph Intelligence Framework for Cyber Physical Smart Grid Stability Prediction. International Journal of Computational Intelligence Systems. 2026;19(1). doi:10.1007/s44196-026-01547-x
BibTeX
@article{Susan2026,
author = {Susan, T. Siron Anita and Marshell, M. Joe and Chandra, B. and Pandi, M. and Rajagopal, S. and Babu, C. Nelson Kennedy and Li, Jinsong},
title = {Energy-Efficient Explainable Graph Intelligence Framework for Cyber Physical Smart Grid Stability Prediction},
journal = {International Journal of Computational Intelligence Systems},
year = {2026},
volume = {19},
number = {1},
doi = {10.1007/s44196-026-01547-x}
}
EUVIMED