Islanding Classification Mechanism for Grid-Connected Photovoltaic Systems

This article develops an islanding classification technique by adapting signal processing and machine learning techniques. The proposed method trains with all the possible islanding conditions, by extracting their features and classifying them. The performance of the proposed method was tested on a single-phase grid-connected photovoltaic system simulated using MATLAB/Simulink environment. The classifier achieved 98.1% training and 97.8% testing efficiency and can effectively detect islanding under 0.2 s with low misclassification. Further, the developed algorithm is tested with a 10-kW grid-connected photovoltaic system to monitor the changes in voltage and power mismatch at the point of common coupling (PCC) and classify the state of the system efficiently.