Research On Blind Source Separation Method Of Bearing Vibration Signal Based On CVMD Decomposition

Rolling bearings are prone to composite failures,blind source separation has achieved some successful applications in the extraction of rolling bearing coupling fault features.However,in practical applications,the problem of"underdetermined" is often faced.Aiming at this problem,this paper studies the application of complex variable mode decomposition in multi-fault vibration signals,and verifies the effectiveness of the method through simulation and experimental research.The main contents of the paper are as follows:Aiming at the problem that the number of sources of blind source separation is difficult to determine,the method of estimating the number of sources based on density is studied.The method is based on the density of the data in space,and the cluster structure is identified by sorting the data points.The basic principle is to set the minimum number of points to find the core point,and sort all the points in the neighborhood of the core point according to the density reachable distance until all the points are processed,and finally obtain the number of cluster clusters.Simulation signals and bearing fault simulation experiments verify the effectiveness of the proposed method.Aiming at the problem of underdetermined blind source separation under actual working conditions,complex variable mode decomposition(CVMD)is applied to the rolling bearing signal,and multiple finite.bandwidth modal components are obtained in the complex domain.The modal component with the same number of source signals is selected by the cross-correlation coefficient,and it is used as the input of independent component analysis(ICA)to realize the separation,feature extraction and diagnosis of bearing signals.Simulation and bearing fault simulation experiments have verified that the CVMD method has shorter running time,higher accuracy and better separation effect than the variable-mode modal decomposition(VMD)method.Aiming at the problem that the selection of initial parameters has a great influence on the decomposition effect of CVMD,a parameter optimization CVMD method is proposed,and a population intelligent optimization algorithm-Drosophila algorithm is introduced.The peak factor of the envelope spectrum is used as the fitness function to search for the optimal decomposition number K and Punishment factor.After determining the optimal number of decompositions and the penalty factor,the composite signal is blindly separated by CVMD and ICA algorithm,and the fault source signal is found and diagnosed.Compared with the CVMD method without parameter optimization,the parameter-optimized CVMD separation effect is more obvious.