Fault Detection in DC Microgrids using Recurrent Neural Networks

Reliable and accurate fault detection plays a crucial role in the microgrid operation by enabling an increased operational flexibility. Successful classification of events in complex microgrid systems requires advanced models of sufficient speed and accuracy. Deep neural networks meet these requirements, as they have demonstrated their capabilities in a wide range of applications. In particular, Recurrent Neural Networks (RNNs) are used for sequence learning, making them suitable for online fault detection. In this work, the RNN is applied to the time-domain signal to detect faults in a photovoltaic-based DC microgrid. The classifier successfully discriminates all events and proves its performance using various metrics.