Reliable protection scheme for PV integrated microgrid using an ensemble classifier approach with real-time validation

The need for enhancing grid resilience has led to wider acceptance of photo voltaic (PV) integrated microgrid. The variation in fault current during grid connected and islanding operation makes the microgrid protection task challenging. A protection scheme based on ensemble of classifiers has been devised by exploiting the effectiveness of a classifier set coupled with voting strategy. Unlike the existing classifier-based approaches involving single classifier, the ensemble-based approach is insensitive to the biasness of individual classifier and dimension/size of dataset. The proposed scheme is formulated to simultaneously perform the tasks of mode detection, fault detection/classification, section identification and location. The scheme is able to discriminate between faults and power quality disturbances, which avoids unintended false tripping. Along with accuracy, the reliability assessment of proposed scheme has been carried out using two indices, i.e. dependability and security for different faults, operating mode and contingencies. For evaluating the reliability of the fault locator, a stochastic approach (Monte-Carlo simulation) has been adopted. The simulation results justify the applicability of the proposed ensemble classifier-based scheme for complete protection and reliable operation of microgrids with PV penetration. The scheme also has been validated for real-time settings using hardware in loop simulations.