A novel pattern recognition-based plan for islanding detection in distribution systems in the presence of synchronous generator-based distributed generation resources

A pattern recognition-based plan for detecting islanding and non-islanding conditions is presented in this paper. This plan is used in distribution systems in the presence of synchronous generator-based distributed generation resources with doubly fed induction generators (DFIGs), a generating principle widely used in wind turbines. The main idea behind the suggested plan is the measurement of total magnetomotive force (MMF) in the SG air gap less than 2 s based on IEEE std.1547 after the occurrence of islanding and non-islanding events. First, a mathematical morphology (MM) filter was used to lessen the possible noise in a total MMF signal sampled in SGs. Then, spatial features like variation, energy, median, rms and mean values of the output magnitude of MM filter were computed. Finally, to distinguish the islanding condition from the non-islanding conditions, support vector machine (SVM) classifier with radial basis function (RBF) kernel was taught based on the above-mentioned features. Different scenarios were considered in various loading conditions in distribution systems for the occurrence of islanding and non-islanding conditions and the determination of the ability of the suggested plan. The outcomes show the method suggested with the SVM (RBF kernel) classifier has much better results in comparison with some other methods.