MATLAB Simulation of SFS Algorithm for Active Islanding Detection Methods

This paper presents a MATLAB simulation of the Sequential Forward Selection (SFS) algorithm for active islanding detection. The SFS algorithm, a widely-used feature selection technique, systematically filters the most representative features from large feature sets by incrementally adding features that maximize detection performance. In our implementation, we actively select optimal feature combinations using MATLAB's classification learner toolbox and custom scripting to validate algorithm effectiveness through ROC curve analysis and detection accuracy metrics. The simulation involves creating distributed generation models with PWM inverters, implementing frequency/voltage drift detection methods, and testing under IEEE 1547 standard conditions. Through this research, we aim to provide a novel solution for islanding detection while offering practical references for researchers employing SFS algorithms in power system protection studies, including code structure for feature ranking loops and classifier integration.