Cyber attack and fault detection in DC microgrids by designing an event-triggered based-robust algorithm

This paper proposes a novel distributed attack detection framework for large-scale systems (LSSs), with a specific focus on low-voltage direct current microgrids (DC MGs). The architecture integrates two complementary detection modules: an event-triggered (ET) observer for local subsystem monitoring and a set of distributed unknown input (UI) observers for assessing the states of neighboring subsystems. To enhance robustness against disturbances, an adaptive compensation mechanism is incorporated. The framework supports an ET control strategy designed to ensure consensus performance, prevent Zeno behavior, and reduce communication overhead. Additionally, a fault detection method based on the state observer is introduced to identify faults within subsystems in real time. The proposed detection method is validated through detailed simulations that consider process noise, model uncertainties, and multiple attack scenarios, including false data injection, stealth, and replay attacks. Results demonstrate that the integrated detection units significantly improve resilience by identifying attacks that would otherwise remain undetected by standalone modules. The study assumes ideal communication links and bounded model uncertainties. Future work aims to address non-ideal communication conditions, investigate time-varying topologies, and develop autonomous reconfiguration strategies based on plug-and-play control.