Dynamic Anomaly Detection With High-Fidelity Simulators: A Convex Optimization Approach

The main objective of this article is to develop scalable dynamic anomaly detectors with high-fidelity simulators power systems. On the one hand, models in are typically “intractable” if opts describe them a mathematical formulation order apply existing model-based approaches from detection literature. other pure data-driven methods developed primarily machine learning literature neglect our knowledge about underlying dynamics In study, we combine tools these two mainstream data-assisted diagnosis filter utilizing both picked abstract model and also data simulation results simulators. proposed aims achieve desired features: (i) performance robustness respect mismatch; (ii) high scalability. To end, propose tractable (convex) optimization-based reformulation which decisions parameters, information introduces feasible sets, simulator forms function to-be-minimized regarding effect mismatch on performance. validate theoretical results, implement DIgSILENT PowerFactory detect false injection attacks Automatic Generation Control measurements three-area IEEE 39-bus system.