An Ensemble Classifier-based Protection Scheme for Faulty Zone Identification in DC Microgrid

As a direct result of the increasing popularity of DC powered equipment, scientific interest in recent years has focused on DC microgrids. However, adoption of DC microgrids by utilities remains limited due to challenges in building reliable protection systems. In DC microgrids, it is difficult to obtain reliable protection due to the large DC fault current amplitude, fast slew rate, and lack of zero crossings. In addition, the energy generated by non-traditional distributed generators is intermittent and therefore needs to be able to adapt to various climate environments. The approach adopted in this study is to develop an ensemble of decision tree-based protection scheme for DC microgrids in terms of detecting the fault and identifying the faulty zone. The scheme has been developed considering the intermittency of PV-based renewable energy sources. The only information used is local information comprising of the voltage-current signals. The developed scheme exploits the inherent characteristics of wavelet transform to derive features from instantaneous voltage-current signals and ensemble classifier approach for correct prediction of faulty zone.