Machine Learning based Clustering for Identifying Power Quality Events

This paper presents an online power quality disturbance identification technique. The AI technique is based on unsupervised machine learning clustering algorithm and pattern recognition approach. The unsupervised machine learning is presented as a powerful tool for clustering application in order to recognize sag, swell and interruption for power quality applications. The K-means clustering algorithms automatically cluster similar dataset samples together and assign every training samples to its closest centroid. In addition to the training based applications, the classifier just needs training at an earlier stage and finally the classifier guarantee machine-like work all the time, conditional upon his past training. Also, the current ongoing critical theoretical contextual of AI application for society and modern power system is addressed in detail. The validated approach is very accurate, flexible to apply for any power system related problems and reasonably fast in identification online based application.