Fault Diagnosis Of Rolling Bearings And Ball Screw Based On Finite Element Simulation Study

CNC machine is the major equipment of modern industry, which is for the production. With the complex internal structure and high degree of automation which can work precisely, it can meet the various technological requirements and precision. But CNC machine tools always work in the condition of variable load and variable speed, and frequent commutation, it may cause the damage of some parts. Such as Rolling bearing and Ball screw is the core components and one of the most vulnerable parts of the feed system of CNC machine tools. So the research on fault diagnosis about Rolling bearing and Ball screw has massive significance to the normal operation of CNC machine tools.This paper takes the Rolling bearing and Ball screw as an object, and the failure mechanism is considered comprehensively; Moreover, it summarizes the common fault types of the Rolling bearing and Ball screw to build the finite element model of the Rolling bearing, and realizing the explicit dynamics simulation of the Rolling bearing with different fault state. The corresponding acceleration and displacement signal are compared and analyzed, and the attribute of signal is selected reasonably; Furthermore, the loss of fault signal of rolling bearing is analyzed, and the further analysis of the fault signal is researched. Based on the analysis of the temperature field and stress of the assembly parts of the Rolling bearing and Ball screw, sensor placement is selected at the sensitive point of the signal.With the external sensors used to collect the fault signal of Rolling bearing and Ball screw, on the basis of muti-channels acquisition board NIPCI-6143 and the program of LabVIEW, the multi-channels signal acquisition system, which is used to collect acceleration and temperature signal, is established. The integrity of the fault signal of Rolling bearing and Ball screw, which is random is assured. The characteristic parameters with the failure information are extracted through the method of wavelet packet analysis to eliminate redundant information and reduce the loss offault information.The method of KPCA is used to screen and optimize the feature extraction, and to realize dimension reduction and simplification of the fault feature extraction, for the accuracy and speed of the subsequent processing; Three pattern recognition methods, including BP, RBF, SVM, is used to realize feature-level fusion to establish a fault diagnosis system of rolling bearing and ball screw; Finally, quote D-S evidence theory to fuse the result of the three methods,called decision level fusion, so that the accuracy of fault diagnosis can be improved effectively.