Islanding detection classification using artificial neural networks

Islanding detection of distributed generation (DG) has become an integral part of today's distributed energy systems. However, current islanding detection system is mostly passive and does not offer intelligent information to detect power quality issues. Using Artificial Neural Networks (ANN) in the electric field has become very common by using classification and recognition algorithms to classify Islanding condition from Distributed Generation (DG) events. In this paper, we proposed a novel DG event classification technique with three principal stages: 1) parameter selection that had a high probability of showing patterns useful for classification of events. 2) Feature extraction helps the neural network reconstruct the pattern of the parameter over a specific period of the occurrence of an event. 3) Model construction and optimization that are used to accurate the model.