Research On Power Quality Analysis Method Based On Empirical Wavelet Transform

With the continuous construction of smart grids and constant innovation in power technology,various types of power electronic devices have also been widely used.The ratios of nonlinear,impactive and fluctuant load in power systems have continuously increased.As a result,it has caused series of power quality problems and serious harm.At the same time,the steady growth of the national economy and the increasingly precise power equipment have put forward higher requirements for the stability of power supply and the reliability of power quality.Accurately detecting power quality disturbance parameters and identifying different types of power quality disturbances are of great significance for the analysis and management of power quality.This paper applies empirical wavelet transform to power quality analysis to realize harmonic parameter detection and disturbance classification and identification.Aiming at the problem of harmonic detection in power quality analysis,a harmonic detection method based on empirical wavelet transform is proposed.Firstly,spectrum detection of harmonic signals is performed to determine the number of spectral divisions of the empirical wavelet transform;Then the signal is decomposed into several harmonic modes of different frequencies by empirical wavelet transform;Finally,hilbert transform and singular value decomposition are used in each harmonic mode to detect the characteristic parameters of each harmonic component.For different types of harmonic signal simulation experiments,the overall detection error of parameters such as amplitude and frequency is within 1%,which indicates that the proposed detection method can effectively achieve harmonic detection in power systems.It is not only suitable for steady-state harmonic parameters detection,but also suitable for transient harmonic parameters detection.It has high detection accuracy under different noise strength and has good noise robustness.For the problem of disturbance classification in power quality analysis,a disturbance classification method based on empirical wavelet transform is proposed.Firstly,empirical wavelet transform is applied to the disturbance signal to obtain a plurality of different frequency modalities;Then six categorical features are extracted from the mode components,classification features of all the disturbance signa form the original dataset,the original dataset is decomposed into seven single-label datasets based on multi-label classification method;Finally,SVM classifiers are constructed on each single-label dataset to train and test the sample datasets.Through the combination of the classification results of each SVM classifier,the final type of disturbance is determined.Classification results in the classification simulation experiment of different types of disturbances,the overall accuracy of single disturbance classification is above 95%,and the overall classification accuracy of compound disturbances is above 84%,which indicates that the proposed classification method can effectively achieve the classification of power quality disturbances.It is not only suitable for single disturbance classification,but also suitable for multi-disturbances compound classification with strong applicability.It has higher classification accuracy under different noise strength and has good noise robustness.