A computational method based on Gustafson-Kessel fuzzy clustering for a novel islanding detection for grid connected devices and sensors

摘要

Fuzzy Clustering-based (G-K) Gustafson-Kessel is used to create the fuzzy rule-based classifier in a grid connected photovoltaic (PV) system where it is tested using specific features in a grid connected PV inverter for detecting islanding condition. It is detected when harmonic content of voltages at the Point of Common Coupling and inverter increases beyond a threshold value. If islanding is not detected, distribution lines are rendered unsafe. The present study uses G-K fuzzy clustering to categorize islanding and nonislanding incidents. Two features based on Total Harmonic Distortion are extracted and used as inputs for the G-K fuzzy clustering classifier. The proposed technique is tested using nonlinear loads and its performance is verified by simulation using MATLAB Simulink. A hardware test set-up is developed to validate the proposed antiislanding technique and the results obtained are discussed.