Islanding detection of distributed generation and classification of voltage sags/swells using machine learning techniques

This thesis presents innovative approaches for detection, classification and characterization of abnormal events in electricity networks. Due to disturbances and/or faults in electricity networks, the abnormal events are created; one such abnormal event is the formation of power system island containing distributed generating resources and the other is the voltage dips and/or swells. This thesis proposes a Support Vector Machine (SVM) based approach for detection and classification of islanding events in a distribution network embedded with Distributed Generation (DG). Furthermore, two innovative approaches, which include three-phase voltage ellipse method and 3D polarization ellipse technique, are proposed to detect, classify and characterize voltage dips and/or swells in electricity networks.