Classification of Power Quality Disturbances Using Support Vector Machines and Comparing Classification Performance

Abstract

In this study pure sine and five kinds of power quality disturbances (PQD) such as voltage swell, voltage sag, voltage with harmonics, transients and flicker are classified by using wavelet based support vector machines (SVM). The performance of proposed method is evaluated by using real time and synthetic data based on mathematical model. Real time data is obtained from national energy system of Turkey. Synthetic data is acquired by using MATLAB. Additionally performance of SVM is compared with artificial neural network (ANN) and Bayes classifier for same future vector and data. Multi-resolution analysis (MRA) technique of discrete wavelet technique (DWT) and Parseval’s theorem are employed to extract the energy distribution features of signals consisting of PQD. When classification performance of SVM is compared with ANN and Bayes classifier, it’s seen that SVM gives the best result both real time and synthetic data