Prediction of unplanned islanding using an energy based strategy

Abstract

In this study a new index is proposed for power system islanding prediction. The total energy absorbed by coherent synchronous generators during unplanned islanding conditions is formulated as an islanding predictor. Decision tree (DT) algorithm is utilised to extract the information gain of the proposed predictor over the input training samples. A comprehensive list of input scenarios, including island and non‐island conditions is constructed. The rotor angles of synchronous generators are estimated by phasor measurements at generators’ terminals and used for calculating the predictor over the input samples. The results of DT algorithm are rearranged as a set of if–then rules for practical applications. The proposed method is implemented in the 10‐machine 39‐bus New‐England test system.