Research On Fault Diagnosis Of Rolling Bearing Based On Time - Frequency Analysis Method

The fault diagnosis for mechanical equipment is increasingly important, especially the rolling bearings which is the most important, the most common in the rotating machinery. In order to ensure the rotating machinery equipment can run safety, reliably, efficiently, the fault diagnosis for rolling bearings is very important and necessary. Since the characteristics of rolling bearings vibration signal is complex and non-stationary, fault feature extraction is the key and difficult step, and also a research hotspot in the fault diagnosis of rolling bearings. In this paper, the generalized demodulation approach to time-frequency and the local mean decomposition are mainly discussed. At the same time, this paper presents the method of fault diagnosis method of rolling bearings based LMD and LSSVM, also puts forward the method of fault diagnosis method of rolling bearings based GDAT and LSSVM. The research contents of this paper are as follows:(1) This paper investigates the GDAT. In this paper, analyze the theory of the content of GDAT. And through numerical simulation, the superiority of the method in the time-frequency extraction is verified. What’s more, apply this method to the rolling bearings fault diagnosis, establishing a method based on GDAT for the fault diagnosis of rolling bearings. Then do the simulation with experimental data, to verify the superiority of the method when extracting rolling bearings fault feature.(2) The LMD algorithm provides a new way of solving the instantaneous frequency for us, which is the direct method. Compared with the Hilbert transform, calculating the instantaneous frequency using the direct method, not only eliminates the end effect, and ensures the information more complete. However, some problems still exist in the direct method. This paper makes the optimized improvement, making the signal obtained by it more accurate and effective. Finally, put forward the method of fault diagnosis method of rolling bearings based on the LMD direct method.(3) In order to make the fault diagnosis of rolling bearing more rapid and accurate, and let it more intelligent at the same time. In this paper, combined with the methods of machine learning, proposed a method of fault diagnosis for rolling bearing based on LMD and LSSVM, and the method based on GDAT and LSSVM. The experimental results show that:these two methods used in fault classification are both better than the LSSVM.