An entropy-based scheme for protection of DC microgrids

The design of an appropriate scheme for protection of DC microgrids is still a significant challenge. This paper proposes a new method for protection of DC microgrids when subjected to various types of faults and modes of operation. The proposed method shares some features of the differential protection scheme and adopts the Shannon entropy to evaluate the "informational value" of current waveforms in the presence and absence of a fault. It uses the relative entropy of the current waveforms at the two sides of the faulty line to detect and characterize a fault. In the case of failure in the primary protection system, the entropy variations of the current waveforms are used to detect the faulty line. The main features of the proposed method include a) accuracy in a noisy environment with poor data synchronization (due to the delay in communication links) and varying network topology, b) fast operation regardless of the fault type and location, c) ability to detect high-impedance faults, and d) provision of backup protection. The efficiency of the proposed method is demonstrated by comparing the simulation results of several case studies in a typical ring DC network with those obtained using an advanced differential protection scheme.