Classification of Power Quality Disturbances Using  Wavelet Packet Energy Entropy and LS-SVM

Yisheng Hu,Kaicheng Li,Ming Zhang

doi:10.4236/epe.2010.23023

Abstract

The power quality (PQ) signals are traditionally analyzed in the time-domain by skilled engineers. However, PQ disturbances may not always be obvious in the original time-domain signal. Fourier analysis transforms signals into frequency domain, but has the disadvantage that time characteristics will become unobvious. Wavelet analysis, which provides both time and frequency information, can overcome this limitation. In this paper, there were two stages in analyzing PQ signals: feature extraction and disturbances classification. To extract features from PQ signals, wavelet packet transform (WPT) was first applied and feature vectors were constructed from wavelet packet log-energy entropy of different nodes. Least square support vector machines (LS-SVM) was applied to these feature vectors to classify PQ disturbances. Simulation results show that the proposed method possesses high recognition rate, so it is suitable to the monitoring and classifying system for PQ disturbances.

Highlights

The deregulation polices in electric power systems results in the absolute necessity to quantify power quality (PQ)
The identification of PQ disturbances is often based on artificial neural network (ANN) [5], fuzzy method (FL) [6], expert system (ES) [7], support vector machines (SVM) [8], and hidden Markov model (HMM) [9]
Our results indicate that solutions obtained by Least square support vector machines (LS-SVM) training seem to be more robust with a smaller standard error compared to standard ANN training using the same features as inputs

Summary

Introduction

The deregulation polices in electric power systems results in the absolute necessity to quantify power quality (PQ) This fact highlights the need for an effective recognition technique capable of detecting and classifying the PQ disturbances. One of the traditional signal processing techniques called Fourier transform provides information in frequency-domain but it does have limitations. One crucial limitation is that a Fourier coefficient represents a component that lasts for all time. This makes Fourier analysis less suitable for non-stationary signals. Wavelet analysis, which provides both time and frequency information, can overcome this limitation. The combined technique of wavelet packet transform (WPT) and least square support vector machines (LS-SVM) for PQ disturbances recognition is presented.

Feature Extraction Using WPT

LS-SVM

Simulation Analysis

Conclusions