A Decentralized Model-Based Fault Detection and Isolation Scheme for MVDC Shipboard Power Systems

Medium-voltage dc (MVdc) shipboard power systems (SPSs) exert tough challenges to existing protection techniques. To detect and isolate component faults in MVdc SPSs, this article develops a decentralized model-based method. In the proposed method, a protected MVdc SPS is partitioned into protection zones composed of power electronic converters and dc cables. For each protection zone, a bank of linear parameter varying (LPV)-unknown input observers (UIOs) is established to achieve the simultaneous detection and isolation of multiple local component faults. The proposed method solves the major limitations of model-based fault diagnosis methods in terms of computational complexity and susceptibility to flexible system topology. In the verification tests with MATLAB/Simulink and laboratory hardware, the proposed method is effective in recognizing switch failures, short circuits (SCs) in dc-link capacitors, and shunt and series faults in dc cables. Moreover, it exhibits good robustness against nonfault disturbances, measurement noises, and modeling errors.