| In recent years,with intelligent manufacturing strategy promoted,mechanical equipment widely used in petrochemical,aerospace and other major fields,and has become more intelligent and motors,and as a key component of rotating machines,rolling bearing is always in high-speed operation condition for a long period of time,very prone to failure or damage.Therefore,how to use intelligent methods for effective health management of rolling bearing is a hot topic for many scholars.This paper mainly focused on the analysis and processing of vibration signals of rolling bearing.Through different experimental data,the fault types of rolling bearing were judged and the health status was divided.The main research contents are as follows:(1)Firstly,in view of the rolling bearing in the process of operation often mixed with large amount of vibration signals generated by background noise,thus affecting the extraction of effective components,In this paper,the Squirrel Search Algorithm(SSA)was used to find the optimal value of the decomposition layer k and penalty factor α in the variational mode decomposition(VMD),formed the VMD of adaptive signal decomposition method,and then,aiming at the problem of excessive dimension of multi-domain fault feature set,Kernel Principal Component Analysis(KPCA)was used to reduce the feature dimension.(2)Secondly,according to two kinds of classification Support Vector Machine(SVM)to the characteristics of the weigh the pros and cons,choose one strategy of Support Vector Machine(SVM)as the classifier,and then build a rolling bearing based on SSA-SVM health management model,including health assessment model based on the root mean square after cosine similarity calculation into 3σ alarm threshold,different health status samples were divided.(3)Finally,the vibration signal data of rolling bearings measured by CWRU and the UC were used to verify the model constructed above.In terms of signal decomposition,the adaptive VMD was compared with GA-VMD and PSO-VMD,which showed its advantages in signal decomposition and noise reduction.In terms of fault diagnosis and health status evaluation,SSA-SVM was compared with not optimized SVM,PSO-SVM and other models,which showed its superiority in calculation time and accuracy. |
