Classification of Power Quality Disturbance using Stockwell Transform

Abstract

This research aims to target power quality disturbances and achieve precise classification using innovative machine-learning techniques. Effective mitigation is required for these disturbances to impact equipment and system stability. Our method is superior to conventional methods in that it utilizes S-transform for feature extraction due to its simultaneous time-frequency localization. We demonstrate how the S-transform captures intricate details of flow disturbances by incorporating Naive Bayes, Random Forest, and K-Nearest Neighbors algorithms. The accuracy of these algorithms is commendable, but their effectiveness in classifying disorders varies. This study presents a tool that automates issue detection and classification, providing a valuable framework for managing power quality. The implementation of this approach will lead to more reliable and stable power systems, which will address a critical need in power quality analysis.