Improvement Of Micro-Grid Islanding Detection Using Smart Methods Based On Multi-Resolution Analysis And Data Whitening

Abstract

These Islanding of Distributed Generations (DGs) in Micro-grids cause serious problems in operation and management of network. Therefore, proper detection of this event is of great importance. This paper presents a proper, fast and accurate method for detecting Islanding mode in unbalanced micro-grids. The proposed method uses multi-resolution analysis for pre-processing of measured data and using whitening for extracting of proper features. As a result, the accuracy of detection in micro-grid relation to measured noise and calculation window will increase. First, the voltage, current and frequency of DG’s terminal is measured and sampled. This data is pre-processed by multi-resolution analysis of relays and then main features are extracted by whitening. Then, the featured are considered by categorizing part whether these features indicate Islanding mode or not. In order to select the best method the classification methods like decision tree, neural network and support vector machine are compared. The proposed method is tested on an unbalanced micro-grid with DGs, wind and photovoltaic sources. Simulated results verify the accuracy of the method.