Enhanced Fault Detection and Localization in Hybrid AC/DC Microgrids Using Current Entropy

Abstract

Hybrid AC/DC microgrids face significant challenges in the detection, classification, and localization of faults due to the diverse fault characteristics of the AC and DC subgrids, and limited fault-current levels of inverter-interfaced distributed generators (DGs). This article proposes a novel unified framework that leverages differential line entropy (DLE) of superimposed currents, operating in the information domain rather than relying on instantaneous magnitudes, to detect and classify various AC/DC faults. The DLE quantifies sustained deviations in the probability distribution of currents during faults, ensuring immunity to false tripping under non-fault transients and external fault conditions. Fault classification indices based on entropy ratios of phase/pole and ground currents accurately identify fault type and the faulty phase/pole. Finally, a non-iterative, double-ended unified fault localization scheme based on an adaptive pseudo-inverse least squares estimation (LSE) with rank-one updates accurately computes the fault location without convergence issues. The effectiveness of the proposed scheme is evaluated using a modified IEEE 33-bus hybrid microgrid (HMG) testbed in MATLAB/Simulink and validated through Hardware-in-the-Loop (HIL) simulations employing a Real-Time Digital Simulator (RTDS)/RSCAD in conjunction with a dSPACE DS1104 DSP controller.