Variational Modal Decomposition Method And Its Application In Mechanical Fault Diagnosis

Dynamic monitoring and fault diagnosis of mechanical system is of great practical significance to ensure the healthy operation of mechanical equipment,accurate positioning of fault location and diagnosis.From a certain level,the effective extraction and diagnosis of the characteristic information of the mechanical fault vibration signal is still the key problem in the field of contemporary fault diagnosis.As the mechanical equipment is increasingly complex,and the various components are interrelated and tightly coupled,making non-linear,non-stationary features more and more significant.Traditional methods of signal processing,such as wavelet decomposition,empirical mode decomposition,intrinsic time scale decomposition and local mean decomposition,etc.,the above method has been widely used in the field of rotating machinery diagnostics.As a new non-recursive adaptive signal decomposition method,variational modal decomposition(VMD),has the characteristics of high precision and fast convergence compared with traditional empirical modal decomposition(EMD)method.In this paper,the defects of VMD are further improved under the support of the National Key Research and Development Program(No: 2017YFC0805103)and the National Natural Science Foundation of China(No: 51505002).Based on the further research on VMD,the parameters are optimized.Algorithm combined with other methods applied to the rotary machine fault diagnosis.The main research contents and innovative achievements are as follows:1.In order to solve the problem that VMD parameters are difficult to be determined,the VMD influence parameters need to be selected.The improved VMD is compared with the original VMD and EMD.The analysis results show the advantages of optimizing the VMD algorithm.2.Aiming at the problem that the vibration signal characteristic of the rotating machinery is difficult to be extracted effectively,the parameter optimization VMD method is combined with the nonlinear method and applied to the fault diagnosis of the rotating machinery.The experimental data verify that the method is effective in the mechanical fault diagnosis.(1)Aiming at the single channel vibration signal,an adaptive multiscale complexity analysis method based on parameter optimization VMD and composite multiscale fuzzy entropy is applied to the vibration signal feature extraction.Based on the multiscale irreversibility method,the parameter optimization VMD is used to detection of deep-time irreversibility.Aiming at the multi-channel vibration signal,an adaptive multiscale multivariate complexity analysis method is proposed on the basis of improving multivariate multiscale entropy and applied to the fault diagnosis of rotating machinery.(2)Based on t-distributed stochastic neighbor embedding,variable predictive model based class discrimination and variable predictive mode based extreme learning machine are applied to the field of mechanical fault diagnosis,combine parameter optimization VMD,it is applied to the fault diagnosis of rotating machinery.