Detection and Classification of Faults Using Intelligent Approach in Microgrid Network

Microgrids serve as vital for incorporating renewable energy sources into the electricity grid. Classifying faults is difficult because of the reverse power flow, blinding of protection, and sympathetic tripping. Conventional overcurrent protection relays have challenges in microgrids due to the restricted fault current generated by grid-connected renewable energy inverters. The paper proposes an approach for classifying faults in a microgrid to address these problems. To classify the fault types based on the Multilayer Feedforward Neural Network (MLFNN). The implementation of MLFNN uses four feature selections like dphi/dt (differential ROCPAD), dv/dt (differential ROCOV(pu/s)), dp/dt (differential active power change with time (pu/s)) and dq/dt (differential reactive power change with time(pu/s). Using these four features in the proposed MLFNN gives better accuracy. The proposed method is assessed using a 6-bus Radial AC microgrid system. Different test cases were undertaken to locate the fault classification by varying fault resistances in the range of 1 to 100 ohms. Moreover, this proposed work has shown favourable results using other models, such as support vector machines, for fault categorization.