Research Of Fault Diagnosis Of Locomotive Rolling Bearing Based On VMD And MCKD

Rolling bearing is the key component of the locomotive running part,and its operation status directly affects the operational safety of the locomotive.Therefore,it is very important to carry out fault diagnosis research on rolling bearings to ensure locomotive operation stability and safety.This paper takes rolling bearings as the object of study.The research is carried out for bearing fault feature extraction and pattern recognition problem,by analyzing the bearing vibration signal.The main contents are as follows.(1)The basic structure and main fault types of rolling bearings are introduced.The bearing vibration mechanism is analyzed,and the bearing fault characteristic frequency calculation formula is derived.(2)The MCKD algorithm is studied and a parameter-optimized MCKD signal denoising algorithm is proposed.Aiming at the problem of MCKD parameter selection,the determination of its three main parameters is discussed.The filter length of MCKD is optimized by using the two-stage grid search method.The two-stage grid search method is experimentally verified and compared with the PSO algorithm and the grid search method.The result shows that the method has a better denoising effect and a higher operational efficiency.(3)The VMD and EMD are compared by bearing simulation signals,and the VMD has better performance in signal decomposition.The PSO optimization algorithm with Gini index as the fitness function is proposed for the problem of penalty factor and number of components selection of VMD algorithm.(4)The adaptive VMD-MCKD fault feature extraction algorithm is proposed.The signal decomposition is performed using the PSO-optimized VMD.The MCKD is performed on the modal component with the largest Gini index to complete the extraction of bearing fault features.Experimental validation using artificially implanted faulty bearing signals and rolling bearing full life cycle signals effectively extracted the fault features.(5)The SVM algorithm is used for small sample bearing fault mode identification.The PSO algorithm is used to improve the SVM to achieve adaptive parameter selection.The fault samples are preprocessed by the adaptive VMD-MCKD algorithm.The modal components with a high Gini index after VMD decomposition and MCKD noise reduction are selected.The morphological energy entropy of each component is calculated to construct the fault feature vector.Different fault types and different fault depths of rolling bearings were classified by using improved SVM.The results showed that the classification accuracy of morphological energy entropy was higher by comparing the classifier with kurtosis and RMS as the eigenvalue.