Diagnosis of multiple and successive faults in hybrid power systems

Develop anomaly detection and fault diagnosis methods for hybrid dynamical models of power systems that can reliably detect and isolate multiple and successive faults, overcoming the prevailing focus on single-fault scenarios in existing approaches. The target systems combine continuous grid dynamics with discrete events, and the methods should function under hybrid dynamics typical of renewable-rich, converter-dominated networks.

Background

Hybrid system models (e.g., hybrid input/output automata and stochastic hybrid automata) have been applied to fault detection and diagnosis in power systems, often assuming single-fault scenarios and fail-safe components. Recent work introduces probabilistic transitions and filtering methods for battery systems, but these advances do not fully address multiple, closely spaced faults in power grids.

The paper emphasizes the need for real-time anomaly detection and robust isolation in the presence of discrete events and hybrid dynamics. It explicitly notes that current methods remain inadequate for multiple and successive faults, keeping this challenge open in the context of power systems.