An Approach for Assessing the Effectiveness of Multiple-Feature-Based SVM Method for Islanding Detection of Distributed Generation

Abstract

Islanding detection is a critical protection issue, as conventional protection schemes such as vector surge (VS) and rate of change of frequency relays do not guarantee islanding detection for all network conditions. Integration of multiple distributed generation (DG) units of different sizes and technologies into distribution grids makes this issue even more critical. This paper presents a comprehensive analysis of the effectiveness of a new method for islanding detection in DG networks. The proposed method, which is based on multiple features and support vector machine (SVM) classification, has the potential to overcome the limitations of conventional protection schemes. The multifeature-based SVM technique utilizes a set of features generated from numerous set of offline dynamic events simulated under different network contingencies, operating conditions, and power imbalance levels. Parameters (such as voltage, frequency, and rotor angle) showing distinguishable variation during the formation of islanding are selected as features for classification of the events. Features associated with different islanding and nonislanding events are used to train the SVM. The trained SVM is tested on a typical distribution network containing multiple DG units. Simulation results indicate that the proposed method can work effectively with high degree of accuracy under different network contingencies and critical levels of power imbalance that may exist during islanding.