Bagged tree based anti-islanding scheme for multi-DG microgrids

Abstract

Microgrids are becoming a prevalent part of the grid due to its’ numerous advantages on energy management system. Microgrids are perceived as the gate way for systematic inclusion of multiple renewable based distributed generations. One of the key concerns for these types of microgrids is detection of absence of the grid. This paper demonstrates a dual use line relay (DULR) based passive detection technique which uses an intelligent electronic device (IED) platform to confirm the islanding condition. Bagged tree ensemble classifier based on synchrophasor measurements has been used in IED as a decision making algorithm. Assorted range of non-islanding and islanding state of affairs have been considered to train the classifier. The measurements obtained from DULR can be used for further analysis at any level of the smart-grid communication hierarchy. The proposed scheme has been seen as fast, effective, economic, and having zero non-detection zone, compared to some of the existing techniques.