Data-Driven Observer Based Detection for Series Arc Fault in DC Microgrids

Series dc arc fault generates arcing noise that propagates throughout the whole system, making it difficult to identify the fault location. In this article, series arc fault detection and localization are investigated for dc microgrids with distributed generation units (DGUs) using data-driven unknown input observer (UIO). The observer is designed using the data collected from the system without requiring knowledge of the system parameters, and it can be used to detect and locate series arc faults. When a series arc fault occurs, the residual generated by the observer increases, indicating the fault's location. The gains of the observers are tuned properly to enhance the state estimation performance and series arc fault detectability. The proposed UIOs are validated through Simulink simulation and hardware-in-loop real-time simulation tests.