Deep Learning Technique for Recurrence Plot-based Classification of Power Quality Disturbances

Abstract

The classification of power quality disturbances (PQDs) is essential for the stability and reliability of the power system. A method to categorize PQD incidents using a recurrence plot (RP) is developed in this work. RP technique is used to transform 1-D PQD into 2-D graphics. PQD events were produced in compliance with IEEE standard 1159–1995 in both single and multiple forms. The 2-D graphics created using RP is fed to the deep learning architectures: Googlenet, ResNet-50 and Alexnet. The features obtained from deep learning were classified using support vector machine, which shows the correct classification of 15 classes with 99.63% accuracy.