Answers are grounded in this paper and cite their sources.
With the increasing deployment of smart energy systems (SESs), cybersecurity vulnerabilities, particularly false-data-injection (FDI) attacks, pose significant risks to the reliability and secure operation of integrated energy infrastructures. This study investigates the resilience of a SES integrating electrical, thermal, and hydrogen energy carriers through a three-step framework. First, stealthy FDI attacks are designed to manipulate electrical and hydrogen demand measurements while remaining difficult to identify through conventional approaches. Second, an extreme gradient boosting (XGBoost)-based machine-learning framework, optimized using GridSearchCV, is employed to reconstruct the compromised demand data. Third, the post-defense performance is evaluated using reconstruction, operational, economic, and statistical indicators. The results show that the investigated attacks can introduce demand deviations of up to 25.76%, while even the minimum investigated deviations of 12.24% for hydrogen and 20.72% for electricity lead to measurable changes in the optimized operation of the system. In particular, the attack scenario produced artificial cost deviations of 6.05% and 1.50% for electrical and hydrogen loads, respectively, and disturbed the state-of-charge and energy-flow behavior of the storage systems. After XGBoost-based reconstruction, the corresponding cost deviations were reduced to 4.92% and 0.41%, respectively, and the post-defense operating behavior moved closer to the baseline condition. The proposed framework also achieved substantial improvements in demand reconstruction. For hydrogen demand, MAPE decreased from 25.20% before correction to 5.22% after correction, while RMSE decreased from 30.67 to 6.75 and R² increased from 0.4848 to 0.9750. For electricity demand, MAPE decreased from 25.55% to 15.12%, RMSE decreased from 221.18 to 131.04, and R² increased from 0.4716 to 0.8146. Statistical tests confirmed that these improvements were significant ( p < 0.001). The results demonstrate that the proposed XGBoost-based framework can effectively reconstruct compromised demand measurements and mitigate their adverse operational consequences, thereby enhancing the cyber resilience and reliability of integrated SESs.
No takes yet. Share an insight, caveat, or question.
Rizi et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: