Research And Application Of A Variational Mode Decomposition Optimization Algorithm

There is a lot of hidden information in the time series and the composition is complex,and the signal needs to be decomposed.The commonly used empirical mode decomposition method can decompose the signal adaptively,but it is easy to produce the phenomenon of signal aliasing.Variational Mode Decomposition(VMD)came into being and gradually developed into an important tool for processing complex signals.Therefore,how to construct an adaptive VMD method has increasingly become a problem of concern.Firstly,through the research of the VMD method and the organic combination of LMD and genetic algorithm,an improved variable mode decomposition(IVMD)is proposed in this paper.Combining IVMD with Extreme Learning Machine(ELM)method,an IVMD-ELM prediction method is proposed.At the same time,the convergence of IVMD and the combined prediction model is demonstrated.Finally,multiple sets of test functions and measured diesel engine operating noise data and acceleration data are selected to verify the feature extraction ability of IVMD,and the prediction model is experimentally analyzed.The errors under different methods are compared,and the rationality and effectiveness of IVMD and IVMDELM are verified.