Research On Power Quality Disturbance Detection And Identification

In recent years,with the rapid development of modern industrial technology,because of the power electronics technology has developed more and more new rapidly and the increasing connection to the grid with energy equipment and the impact loads,the grid power quality disturbance problem is much more serious than before,it also brought a series of problems to the national production and life,which caused huge economic losses.To solve this problem,it is necessary to analyze and govern the disturbance of power quality.And the key to improve the electric quality,is detecting the power quality problems faster and more accurately,more over,identifying it’s types.In this paper,we study the detection and identification of power quality respectively.The detection and identification of the existing methods,particularly the S transform principle and calculation method are analyzed.Bass on these,the modified S transform(MST)is proposed.In the detection area,we realize the detection of of power quality disturbances such as harmonics,voltage sags,and voltage sag by modified S transform.The comparative experiment with S transform show that the modified S transform based detection method is more effective and more accurate.In identification of power quality disturbance,firstly,we analyze 7 kinds of disturbance i.e.harmonic,swell and sag,oscillation,pulse,harmonic and swell harmonic sag by using the modified S transform.Choosing the amplitude envelope curve,improved extraction S transform on the signal fundamental frequency amplitude curve and time square and three amplitude curve curve as characteristic curve.Then,two kinds of methods are used to classify them.One method is using the support vector machine(SVM)based on the two classification to identify the disturbance above,then we construct the support vector machine tree of the two classification;Secondly,extreme learning machine(ELM)is used,we characterize three kinds of characteristic curves with mean,standard deviation,skewness,kurtosis and RMS values.And obtained 15 features as the input of extreme learning machine,in order to obtain good classification effect.The classification results mentioned above proves that the effectiveness of the power quality disturbance feature extracted by the modified S transform is effective.