Unknown Input Observer-Based Series DC Arc Fault Detection in DC Microgrids

In this article, a fault detection and isolation technique for series arc faults in dc microgrids with multiple power electronics loads is proposed using unknown input observers (UIOs). The dc microgrid is first modeled as a linear system using its corresponding graph information, including the oriented incidence matrix. The lines are assumed to be in Pi form (resistive, inductive, and capacitive) defining the dynamic states: line currents and node voltages. Sources and loads are considered inputs to the network and details are presented to include the case where these are not measured. DC series arc faults are then analyzed as disturbances affecting any line. An UIO framework is then presented to detect and identify the faulted line using the network model. Simulation and experimental results are then presented.