A New Approach for Fault Classification in Microgrids Using Optimal Wavelet Functions Matching Pursuit

This paper introduces a new approach that uses a combination of wavelet functions and machine learning for fault classification in microgrids (MGs). Particle swarm optimization is applied to identify the optimal wavelet functions combination that serves as a matching pursuit to extract the most prominent features, which are hidden in the current/voltage waveforms when applying the discrete wavelet transform. Four different classification techniques (i.e., decision tree, ${K}$ -nearest neighbor, support vector machine, and Naïve Bayes) are used to automate the procedure of fault classification in MGs and their performances are statistically compared. The consortium for electric reliability technology solutions (CERTS) MG is used to exemplify the effectiveness of the proposed approach after modeling the MG system in power systems computer aided design/electromagnetic transient direct current (PSCAD/EMTDC) software package. The results are presented, discussed, and conclusions are drawn.