Detection and Mitigation of Electrical Faults in Medium Voltage DC (MVDC) Architectures

The proposed technology has the potential to change the paradigm such that the traditional “brute-force” method of the mechanical breaker is replaced with natural commutation. Further, a downfall of the conventional solid-state breaker is that it still must interrupt large fault currents requiring very specialized semiconductors and controls. Ultimately, the dc breaker herein prevents large currents from developing across the fault while completely isolating the source bus including the solid-state switch without the use of sophisticated controls. Preliminary real time simulation studies would address the proposed breaker operation within an MVDC shipboard power system. Information from these studies would be used to develop an intelligent network designed to automatically optimize the distribution configuration. This network will be developed using a neural network approach based on supervised and unsupervised learning, such as self-organizing maps and learning vector quantization networks.