Investigation On Fault Features Enhancement Method And Condition Prediction For Rolling Bearing

Rolling bearing is a key component of the machinery device. It has an important impact to the whole unit whether its working condition is normal. Rolling element bearing fault feature is very weak in the incipient process and interfered easily by normal features and noise on the spot, and this brings much difficult to actual diagnosis, so how to diagnose rolling bearing fault actually in the early stage is an urgent problem. Besides, we usually can't and don't need to stop running for maintenance or replacement when find incipient fault. The normal measure is to take regularly monitoring and forecast reasonably and actually according to historical data, so forecast is an important part for machinery equipment preventative maintenance. These two questions are investigated separately in this research and put forward some effective measures.Wavelet packet decomposition is one effective method and is commonly used in rolling bearing fault diagnosis. Frequency bands obtained by wavelet packet decomposition are investigated in this research. A new fault diagnosis method is put forward based on wavelet packet-coordinate transformation (WP-CT) for feature enhancement. As every frequency band obtained by wavelet packet containing normal or fault feature, principal component analysis (PCA) or independent component analysis (ICA) is used for analysis. Then, a new signal with fault feature can be reconstructed. Simulated signal is used to testify the effectiveness of this method. It can be concluded that this new WP-CT method can enhance fault feature and reduce the interference from normal signal, which is very helpful for rolling bearing fault diagnosis technology development.As for the condition forecast of rolling bearing, GM (1,1) model whose modeling mechanism is very simple is selected in this research. GM (1,1) model is the most respective basic content of the gray theory. Its modeling feature of little information, high precision and simple characteristic has been used widely in many fields. However, traditional GM (1,1) model still has some defects, for example, prediction error is large, and applications scope is narrow and so on. GM (1,1) model is investigated in this research in order to improve its practicality in the practical predictive maintenance equipment and then a new grey improved algorithm is put forward based on weighted and minimum average relative error. According to comparison, the accuracy of this new algorithm is higher and the scope of application is wider. At the same time, this method is applied to practical condition prediction of rolling element bearing and achieves good results. Besides, the modified algorithm is helpful for machinery equipment preventative maintenance according to the results analysis.