Research On Identification Of Voltage Sag Sources Based On IVMD And M-SVM

With the acceleration of modern industrialization,digital automatic control technology based on frequency converters and programmable logic controllers is widely used in industrial production.These devices are very sensitive to voltage changes,therefore,greater demands are being placed on the reliability of power quality.As a common transient power quality problem,the fast and accurate identification is an important prerequisite for analyzing,suppressing and compensating the voltage sag problem.In order to identify the voltage sag sources reliably,the causes of different types of voltage sag and their sag waveforms are systematically studied.By comparing several common detection algorithms and their advantages,the variational mode decomposition(VMD)method is introduced into the detection of the voltage sag sources.However,no clear principle is set up for VMD parameter configuration,mainly the decomposition components K and equilibrium parameter ?.An improved variational mode decomposition(IVMD)method is proposed,the number of VMD components is determined by short time Fourier transform,and the approximate fitting formula of the best balancing parameter ? is obtained by a large number of tests.The voltage sag signal is decomposed into several band-limited intrinsic mode functions,and calculated the multi-scale entropy of each component as feature vector,the principal component analysis is used to reduce its dimension as a final feature vector.Multi-level support machine combined with improved particle swarm optimization algorithm is used to construct a multi-classifier to realize accurate identification of voltage sag sources.The voltage sag simulation system model is built in PSCAD/EMTDC software,and the voltage sag sample data of different types of disturbance sources are obtained.The method proposed in this thesis is verified.The simulation results show that the proposed method has high accuracy and reliability for feature extraction and identification of voltage sag sources.The thesis has 37 pictures,5 tables and 55 references.