Islanding detection for PV and DFIG using decision tree and AdaBoost algorithm

Under smart grid environment, islanding detection plays an important role in reliable operation of distributed generation (DG) units. In this paper an intelligent-based islanding detection algorithm for PV and DFIG units is proposed. Decision tree algorithm is used to classify islanding detection instances. This algorithm is rapid, simple, intelligible and easy to interpret. The error rate of this method is reduced by Adaptive Boosting (AdaBoost) technique. The proposed method is tested on a distribution system including PV, DFIG and synchronous generator. Probable events in the system are simulated under diverse operating states to generate classification data set. First and second order derivatives of locally measured electrical parameters are used for construction of 16-dimensional instances. The results indicate that Adaboost technique yields improved islanding detection accuracy. This algorithm is capable of detecting islanding phenomenon under operating states with negligible power mismatch.