Microgrid security assessment and islanding control by Support Vector Machine

A Support Vector Machine (SVM) based approach for microgrid islanding decision and control is investigated. The IEEE 13-feeder system is modified and serves as the microgrid model connected to Kundur four-machine two-area system that models the main transmission grid. A representative data set is obtained through simulations in MATLAB/Simulink considering multiple typical scenarios with or without a fault. A SVM classifier is designed to identify insecure scenarios with satisfying accuracy. Comparisons between different kernel functions are then carried out, which indicate that linear SVM can be effective for the islanding control. The SVM approach is further compared with a decision tree based approach in terms of training and testing accuracies for the microgrid islanding control problem.