Power Quality Detection And Simulation Analysis Based On Wavelet Transform

In recent years, power quality has attracted more and more attentions by electricpower department and its users. Power quality detection is a necessary prerequisiteto monitor and improve the power quality, which has important theoretical andpractical significance to ensure the safe and economic operation of power systems.This paper focuses on the time locating and classification of common power qualitydisturbance signals.Firstly, this paper gave an overview of research of power quality detectiondevelopment at home and abroad; described the definition and classification ofpower quality from different angles; analyzed and summarized the relevant nationalstandards of power quality and the development trend of power quality detection;gave7mathematical models of common power quality disturbances. Secondly,wavelet theory and its natures were introduced in detail, and its application in powerquality detection was discussed. It was mainly focused on principles of the powerquality disturbance singularity signals detection and extraction methods ofclassification feature vector based on wavelet transform. Through the simulationanalysis, it exhibited the distinguished space of the extracted feature vector inthree-dimensional perspective, and verified the validity of the extracted classifiedfeature vectors.Detection and localization of power quality disturbance signals provide thebasis for analyzing the causes of the disturbance, this is important significance. Thispaper presented a complex wavelet-based power quality disturbance detection andlocalization. First of all, it extracted the amplitude and phase of complex waveletcoefficients through discrete complex wavelet transforming; then, used compositeinformation of amplitude and phase to achieve time positioning of5kinds oftransient power quality disturbance signals. The method is still useful in the noiseconditions. However, the method will fail in short duration power qualitydisturbances when the starting and ending points had occurred in the vicinity of thezero amplitude of signals. In this case, this paper proposes a new assistedpositioning methods, which decoposited and reconstructed wavelet to get lowfrequency signal and do complex wavelet transform. Simulation shows that under thecondition of noise the method can pinpoint quickly power quality disturbance signals.Precise identification and classification for power quality disturbances issignificantly important to analyze and comprehensively cope with power qualityproblems. Based on wavelet and improved neural tree techniques, a newclassification methodology for power quality disturbances is proposed. In themethod, the disturbance signal is decomposed into different frequency bands, whilstenergy values and wavelet coefficient entropies of the base, harmonic and highfrequency bands are calculated as eigenvalues respectively. The root mean producedin the disturbance process of the base wave band was calculated as a supplement,which was then combined with the energy values and wavelet coefficient entropiesas eigenvectors for judging the disturbances. Thereafter the eigenvectors werenormalized and inputted into the improved neural tree classifier, composed of neuralnetwork, decision trees and classification rules, for training and classifying.Simulation results demonstrate the method has a small amount of calculation toextract eigenvalues and the obtained eigenvectors can adequately reflect thedifference information for different disturbance signals. The improved neural treeclassifier combines respective superiorities of the neural network and decision treein pattern classification, thus the classifier presents good convergence, globaloptimality and generalization, and can effectively identify7common power qualitydisturbances with a simple structure and high accuracy.