Islanding Detection of Microgrid using EMD and Random Forest Classifier

This article focuses on islanding detection of a distributed generation (DG) based micro-grid system. The presented scheme operates in a two-stage process. First, different features of the system are extracted using empirical mode decomposition (EMD) under certain conditions of islanding, capacitor switching and load switching. Second, a random forest (RF) is trained and tested with several rows of dataset so as to predict the class of events accurately. The system is designed and simulated under MATLAB simulink environment and it is found the discrimination capability of RF is better than the competing machine learning tools like Decision Tree (DT), Support Vector Machine (SVM) and Naïve Bayesian Classifier (NBC).