Fault zone identification and phase selection for microgrids using decision trees ensemble

Abstract Microgrid protection is one of the main challenges that hinders the growth in the number and variety of microgrids being developed. Fault zone identification and phase selection are among protection challenges that existing protection schemes fail to address. This paper proposes a voltage-restrained classifier-based relaying scheme to address the two protection selectivity problems. While fault detection is achieved via an undervoltage function to ensure the sensitivity of the proposed method, Decision Trees (DTs) ensemble functions are used for fault zone identification and phase selection to guarantee the selectivity of the protection scheme. The proposed approach can identify three zones, forward internal, forward external, and reverse external, which (i) facilitates the operation of a relay in primary and remote backup modes, and (ii) avoids nuisance operation under reverse faults, i.e., works as a directional element. The proposed phase selection function can identify all ten types of faults, which enables the isolation of the faulted phase(s) only. The proposed relaying logic does not require any form of communication, instead it relies on readily available sequence component measurements. Transient studies in PSCAD/EMTDC verify the performance of the proposed scheme under various faults and different microgrid operation modes.