Date of Award
August 2026
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Electrical Engineering
First Advisor
Naima Kaabouch
Abstract
Modern cyber-physical power grids face unprecedented threats from Coordinated Cyber-Physical Attacks (CCPAs). Current data-driven defense mechanisms frequently suffer from data scarcity, lack adherence to physical power system laws, and operate in isolated silos, leaving them structurally blind to sophisticated intrusions such as Relay Setting Cyber-Attacks (RSCAs) and Wide Area False Data Injection Attacks (FDIAs). To address these critical vulnerabilities, this dissertation proposes a unified two-tier, dual-layer physics-informed cyber-resilient protection architecture. First, to overcome the data scarcity bottleneck, a high fidelity, closed-loop co-simulation testbed is developed utilizing HELICS, ATP-EMTP, MATLAB, Python, and the ACTIVSg200 synthetic grid, generating comprehensive datasets for both electromagnetic transient (EMT) and synchrophasor domains. Second, at the substation tier, a Dual-Relay Temporal Physics-Informed Neural Network (DR-TPINN) is introduced. Supervised by differentiable electromagnetic consistency residual losses, the DR-TPINN detects local RSCAs directly from process bus waveforms. Third, to evaluate wide area vulnerabilities, a Graph Wasserstein Generative Adversarial Network with Gradient Penalty (GraphWGAN-GP) is deployed to synthesize mathematically stealthy FDIAs capable of bypassing traditional bad data detection. To neutralize these wide area threats, the control center tier employs a Physics-Informed Graph Neural Network Dynamic State Estimator (PI-GNN-DSE) that integrates Extended Kalman Filter priors into a multi-head Graph Attention Network. Finally, the proposed architecture fuses the substation tier's misoperation score with the control-center tier's innovation score. By closing the cyber-physical observability gap, this unified framework actively rejects adversarial data, prevents malicious topological masking, and provides operators with a comprehensive, cyber-resilient operational view of the grid.
Recommended Citation
Caleb, Toro Dama, "Physics-Informed Cyber-Resilient Power System Protection Scheme Against Coordinated Cyber-Physical Attacks" (2026). Theses and Dissertations. 13865.
https://commons.und.edu/theses/13865