Rolling bearings are the most conventional components in vast majority of rotating machines, and they are also easily to be damaged. Nowadays, in most companies the quality inspection of the rolling bearings are mostly based on the personal sense of QA Test Technician, therefore, it is very hard to eliminate individual subjective factors, and the result will not be very good. To solve these problems, in this paper we research the methods of the fault detection of rolling bearings based on vibration signal processing. The specific work are maily as follows: Firstly, the vibration mechanism of rolling bearing structure, the failure forms, vibration types and source of trouble are analyzed. The formula of characteristic frequency of fault bearing is given. It provides theoretical foundation for the bearing fault diagnosis. Secondly, the application of the envelope spectrum of wavelet analysis is studied. Threshold de-noising method in wavelet domain is used for bearing vibration signal denoising, which forms the foundation for the analysis of the bearing fault signal. By the methods of wavelet decomposition, reconstruction and hilbert transform, the characteristic frequency of fault bearings is picked up, and it would be compared with the theoretical frequencies. Then, which kind of faults of the rolling bearings can be determined Finally, a large number of experimental data are analyzed, and the result is satisfied. It can be proved that the method used in this paper is correct and it also means that the wavelet analysis theory is one of the most useful methods for faults diagnosis of rolling bearings. An acceleration sensor data acquisition system is designed here based on single-chip microcomputer.
修改评论