A Centralized Protection Scheme for Microgrids with Artificial Neural Network-Based on Fault Detection and Location

The increased deployment of renewable energy sources (RES) as distributed generation (DG) in power grids brings a challenge to the protection scheme. Due to different penetration levels of RES during a day, the fault currents at the same point of the microgrid (MG) vary significantly. Since, the MG presents two different levels of fault current according to grid-tied mode or islanded mode. Consequently, the conventional overcurrent coordination protection schemes must be developed. This paper proposes an improved centralized protection strategy for AC MG with bulky DG penetration. The proposed strategy depends on communication-based overcurrent relays with a centralized unit using an artificial neural network (ANN) with symmetrical components as feature extraction. The proposed algorithm provides fast fault detection and fault location. The evaluation of the proposed strategy is validated on IEEE 9-bus system using Matlab/Simulink software. The system is examined under different operating conditions of MG and different fault types at different fault resistance and achieved remarkable results.