Skip to Main Content
The proposed method develops a fuzzy rule-based classifier that was tested using features for islanding detection in distributed generation. In the developed technique, the initial classification boundaries are found out by using the decision tree (DT). From the DT classification boundaries, the fuzzy membership functions (MFs) are developed and the corresponding rule base is formulated for islanding detection. But some of the fuzzy MFs are merged based upon similarity the measure for reducing the fuzzy MFs and simplifying the fuzzy rule base to make it more transparent. The developed fuzzy rule-based classifier is tested using features with noise up to a signal-to-noise ratio of 20 dB and provides classification results without misdetection, which shows the robustness of the proposed approach for islanding detection for distributed generations in the distribution network.