Toward intelligent fault classification in autonomous microgrids

A fault detection method for an inverter-based microgrid is proposed. This microgrid consists of inverters, motors, and other loads that increase the probability of fault events. Line-to-line inverter faults and induction motor faults are analyzed and their detection methods are discussed. Sequence networks and FFT analysis are used for feature extraction, to be used as input to the artificial neural network (ANNs). The multilayer perceptron ANNs have then been used for diagnosis purposes. Simulation results validate model accuracy for fault detection of faults and localization.