Research On Power Quality Disturbance Detection And Analysis Based On EEMD And SVM

With the rapid development of social economy, power electronics technology is also increasingly applied to the production and living among us, a large number of nonlinear and impact load access to power grid, making the grid voltage waveform distortion, even lead to frequency fluctuations throughout the distribution system. At the same time, the microprocessor as the core of electronic equipment and precision instruments of a higher power quality requirements. Therefore, power quality issues more and more attention and research scholars of the power sector.The main innovation and research work is as follows:1) Using mathematical models of power quality disturbances, analysis and study of several major factors that affect power quality. Through overall average empirical mode decomposition (Ensemble Empirical Mode Decomposition, EEMD) and Teager energy operator (Teager Energy Operator, TEO) on the extraction of disturbance signal decomposition and energy characteristics, can detect the occurrence of the phenomenon of power quality disturbances, and access to power quality disturbance energy feature data. Based on the above energy characteristic, according to some factors affecting the quality of electric power, the support vector machine (Support Vector Machine, SVM) to classify and recognize the disturbance type.2) Considering the mode function serious mixed phenomenon in empirical mode decomposition (Empirical Mode Decomposition, EMD) method, with the overall average empirical mode decomposition method, for anti-aliasing the modal function obtained by the decomposition of the voltage signal, and then use the Teager energy operator gets the energy characteristics of disturbance signal, according to the disturbance signal energy feature calculation get, can detect power quality disturbance phenomenon can occur.3) In the classification of disturbance signal, according to the artificial neural network classifier, Bias and other existing problems in power quality disturbance signal in the classification, support vector machine is introduced to the power quality disturbance identification. Through the comparative analysis of the radial basis kernel function and polynomial kernel function and the linear kernel function of support vector machine classification recognition disturbance effects, using PSO algorithm and genetic algorithm to optimize the radial basis kernel function optimal parameters, and disturbance energy characteristics of the signal as the support vector machine. The input vector is applied to classify and identify the signal.The simulation results have proved the effectiveness of this method, the overall average empirical mode decomposition accuracy higher than the empirical mode decomposition, the disturbance signal detection process reflects the rapidity and accuracy of the method. Classification and recognition process, using the parameter optimization method for further demonstrates the reliability research methods in the thesis.