Title: An efficient islanding detection method in distributed generation using hybrid SVM-based decision tree
Authors: Gundala Srinivasa Rao; Gattu Keseva Rao
Addresses: Department of Electrical and Electronics Engineering, KL University, Vaddeswaram, Guntur, India ' Department of Electrical and Electronics Engineering, KL University, Vaddeswaram, Guntur, India
Abstract: Islanding is one of the most important concerns of interconnecting the grid-connected distributed resources to the distribution system. At the point when a bit of the distribution system turns out to be electrically detached from the rest of the power system, yet keeps on being empowered by distributed generator (DG) islanding happens. Islanding is an undesirable circumstance, since it is conceivably a hazardous condition for the upkeep work force and might harm the DG and loads on account of unsynchronised reconnection of the lattice because of stage distinction between the grid and DG. So effective and accurate islanding detection is essential to protect the distributed system while landing occurs in a distributed network. In this paper, a hybrid support vector machine with decision-tree classifier is proposed to provide an accurate detection and classification of islanding based on extracted features within less detection time. The proposed method is actualised in MATLAB, and the test results demonstrate the significance and viability of our proposed system than the current islanding discovery strategies.
Keywords: DG; distributed generator; forward feature selection algorithm; islanding; support vector machine with decision tree.
International Journal of Power Electronics, 2018 Vol.9 No.2, pp.189 - 213
Received: 20 Sep 2016
Accepted: 15 May 2017
Published online: 28 Sep 2017 *