Search

Cyber-physical attacks on smart grids can compromise the measurements, forecasts, and control signals used in dispatch. A resilient decision-support system must detect anomalies, explain their operational meaning, and support secure re-dispatch before manipulated data causes unsafe operation. This article proposes a cyber-resilient decision-support system that integrates anomaly detection, explainable alerts, and robust optimization for smart-grid dispatch. The system is designed as an operator-facing architecture rather than a standalone detection model. The system combines machine learning-based anomaly detection, explainable artificial intelligence-driven alert generation, and robust dispatch re-optimization. Validation is structured around false data injection, voltage measurement manipulation, and load manipulation attacks on an IEEE test feeder. The proposed decision-support system identifies cyber-physical anomalies, converts model outputs into actionable explanations, and updates dispatch decisions under attack-induced uncertainty. Compared with non-resilient dispatch and detection-only benchmarks, it reduces unsafe voltage excursions and avoidable line-flow stress. The integrated architecture improves cyber-resilience by connecting detection, explanation, and mitigation in one operational workflow. It supports faster operator interpretation while maintaining dispatch feasibility under uncertain measurements.

The shift of electrical power systems toward intelligent smart-grid architectures delivers better dependability, operational performance, and real-time visibility, yet simultaneously heightens exposure to cyber threats. Thanks to the massive amounts of data now generated in smart grids, machine learning techniques have emerged as a particularly promising avenue for bolstering defensive capabilities. While the strengths of machine-learning solutions for cybersecurity improvement are clear, these same models form part of the attack surface and remain especially susceptible to carefully crafted adversarial perturbations. This work analyzes the resilience of autoencoder-based intrusion detection mechanisms in smart grids against adversarial examples. We begin by presenting an original iterative procedure for synthesizing adversarial instances. Next, we demonstrate that a white-box adversary with full knowledge of the autoencoder detector can reliably produce stealthy, misclassifying samples using the proposed approach. Experimental results show that simply starting from basic adversarial seeds is insufficient to generate effective evasions, underscoring the considerable difficulty of constructing potent adversarial examples targeting autoencoder-based anomaly detectors deployed in smart-grid environments.