Power Quality Disturbance Identification Based On Variational Mode Decomposition And Machine Learning

With the continuous deepening of smart grid construction and the increasing maturity of new energy grid-connected technology,great changes have taken place in the composition of power sources and loads.The grid connection of distributed and intermittent power sources,such as wind power and photovoltaic power generation,brings a series of power quality problems to the power system,such as harmonic pollution,transient oscillation and so on.Therefore,the problem of power quality has become a common concern of the power supply department and the majority of users.The premise of solving the power quality problem is to identify the power quality disturbance signal effectively.Therefore,this paper studies the problem of power quality disturbance identification from two aspects: feature extraction and pattern recognition.In the aspect of feature extraction,this paper uses the method based on improved Hilbert transform(HHT).In this method,the empirical mode decomposition(EMD)which is part of the original HHT transform is improved by variational mode decomposition(VMD),and the effective time-frequency information is extracted by making use of the energy sensitivity of marginal spectrum.In the aspect of pattern recognition,the models of two kinds of power quality disturbance identification are constructed in this paper.(1)The model of power quality disturbance identification based on VMD-PCA-SVM.In this paper,support vector machine(SVM)and principal component analysis(PCA)are organically combined to build a PCA-SVM model.We firstly generate a batch of disturbance signals simulated by Matlab2019 a and then add white Gaussian noise.Secondly,it is convenient to extract the marginal spectrum uesed by the improved HHT algorithm and reduce its dimensions used by PCA.Last but not least,it is inputted into SVM to train and test.The experimental results show that under the conditions of signal-to-noise ratio(SNR)of 30 d B,40 d B and 50 d B,the identification accuracies of VMD-PCA-SVM model for single disturbance and double disturbance are more than 99%;under the condition of SNR=20 d B,the identification accuracy for double disturbance is also more than96.5%,which shows its good accuracy and robustness.(2)The model of power quality disturbance identification based on VMD-DBN-ELM.In this paper,a deep confidence limit learning machine(DBN-ELM)model is constructed by combining extreme learning machine(ELM)with deep confidence network(DBN).An improved HHT algorithm is used to identify power quality disturbances.The experimental results show that the model also has good accuracy and robustness.Under the condition of SNR=20 d B,the accuracy of double disturbance identification is 1.0% higher than that of VMD-PCA-SVM.The models of two power quality disturbance identification proposed in this paper are the application and extension of VMD and machine learning in the field of power quality disturbance identification.The simulation results show that the two models are effective and provide two effective solutions for power quality disturbance identification.