Research On Fault Diagnosis For Rolling Bearing Based On Variational Mode Decomposition

With the process of industrialization in China,as the aging of the population,the production and operation of the machine have begun to enter the aging period which will reach a large number in the future.Rolling bearing is one of the most important parts of rotating machinery,and it is also one of the biggest failures.Therefore,it is of great practical and economic significance to study the fault diagnosis of rolling bearing.Modal extraction is the key to the fault diagnosis of rolling bearing,especially for the fault feature extraction of rolling bearing.The quality of feature extraction has a direct impact on the fault diagnosis results because the vibration signal of rolling bearing is typical nonlinear signal.Aiming at the problem of fault feature extraction and recognition,the main contents are as follows:(1)Through the introduction of basic concepts of variational mode decomposition method of intrinsic mode function,which is Wiener filtering and the analytic signal,the problem of VMD method is illustrated to construct signal constraints.And then the variational decomposition method is introduced to solve constrained problem.In order to verify the superiority of the VMD method,the decomposition of the noise interference signal and the pulse interference signal are analyzed by VMD method and empirical mode decomposition method respectively.The results show that the VMD method has obvious advantages in the noise robustness and the impulse disturbance.(2)In order to extract the fault features of the rolling bearing accurately and stably from the signal of the complex bearing in strong noise,a method which is VMD and stationary wavelet method based on kurtosis criterion is used for bearing fault diagnosis.First,use variational modal decomposition under the same load fault signal preprocessing,and then evaluate modal by kurtosis to filter the best and the second best signal to be reconstructed and denoised it by stationary wavelet.Finally,analyze the envelope of the signal spectrum to diagnose the type of the bearing fault.Through the analysis the simulation signal of the inner ring of rolling bearing,the method can extract the information of the weak characteristic frequency from it and the effect of the noise suppression is better than that by EMD.The results show that the fault diagnosis of bearing based on VMD and stationary wavelet can effectively distinguish the incipient fault information of rolling bearing from strong noise,which has certain reliability and application value.(3)In order to realize the fault diagnosis of the rolling bearing accurately,based on variational modal decomposition Instantaneous energy method and support vector machine optimized by mutation particle swarm optimization,a fault diagnosis method is used.First,the fault vibration signal of the rolling bearing is decomposed by variational mode decomposition.Then,according to the characteristics of VMD components which contain the main fault information,the components are selected to calculate the energy feature and build fault feature vector.Finally,feed it into the support vector machine optimized by mutation particle swarm optimization to sort the working state and fault type of rolling bearing.The method can be used to classify the bearing vibration signal more accurately,which has a good classification results.