Voting Technique for Islanding Detection in Microgrid Using Superimposed Angle Variations

The detection of islanding events in a microgrid is a problematic issue. In this paper, four angle variations of superimposed sequence components are derived, and a new technique is proposed using the highest aggregate votes. Space vector transform is applied to decompose positive sequences and negative sequences from power signals. From the pre-state and post-state information, mathematical relationships of superimposed components are generated, and angle variation in between the superimposed component and pre-event component of each sequence is computed. This technique identifies the islanding state, even under perfectly power match conditions without the threshold setting, and hence overcomes the non-detection zone (NDZ) problem. Also, the non-islanding states such as the variable fault resistance, single-pole tripping, recloser operation, non-linear loads, and reconnection of the DEU are addressed. The proposed technique is verified using the thirteen bus microgrid simulation model in a real-time digital environment. The proposed technique is found to be fast and accurate on the change in source capacity, location, and type of DEU.