Fault diagnosis model based on NRS and EEMD for rolling-element bearing
Jin Lian, Rongzhen Zhao · 2017
Considering the fact that the early fault of the rolling-element bearing is difficult to identify correctly, a neighborhood rough set (NRS) and ensemble empirical mode decomposition (EEMD) fault diagnosis model of rolling-element bearing combination is come up with. First of all, the original signal of the vibration is decomposed by EEMD to obtain a number of IMF components, and extracted time-features from the first 3 IMF components in the time domain to form the original features. Then, NRS is used to reduce the attributes of original features, eliminate the redundant information, and put the sensitive reduced feature attributes into the SVM classifier for fault identification. This model is applied to typical rolling-element bearing fault diagnosis experiments, which shows that the NRS is used to select a large number of features containing abundant fault information from the original features. The method not only reduces the complexity of the classification algorithm, but also enhances the accuracy of the fault identification by 5%, which provides a new approach of analysis model for rolling-element bearing fault diagnosis.