Series DC Arc Fault Detection in DC Microgrids Based on Distributed Unknown Input Observers

Series dc arc faults detection in large-scale dc microgrids can be challenging due to high frequency noise crosstalk and large computational load. This paper proposes a distributed Unknown Input Observer (UIO) scheme to detect series arc faults in large-scale dc microgrids. The microgrid network is divided into subsystems and modeled as linear systems using graph information and the oriented incidence matrix. The inputs to each subsystem are loads and sources, and interconnections between neighboring subsystems. The series dc arc fault is modeled as a disturbance that can occur at different lines and subsystems. To detect various line faults, a set of distributed UIOs is developed for each line. An optimization method is then used to compute the observer gain and threshold using linear matrix inequalities. The proposed UIOs are validated through simulation and experimental case studies. The results demonstrate that the proposed filters can effectively detect and locate dc arc faults in large-scale dc microgrids with multiple zones.