Fault Diagnosis based on Rough Set and Dependent Feature Vector for Rolling Element Bearings
Download citation file:
The fault diagnosis of rolling element bearings (REB) has attracted substantial attention recently due to its importance for the bearing health management. The methods based on empirical mode decomposition and intelligent classification are widely used for REB fault diagnosis. However, there still exists two shortcomings in the fault diagnosis methods: 1) A large amount of redundant information is difficult to identify and delete; and 2) Aliasing patterns decreased the classification accuracy. To deal with these two shortcomings, an improved fault diagnosis method based on rough set and dependent feature vector (RS-DFV) is proposed in this paper. In RS-DFV method, the dependent feature vector (DFV) is used to optimize the feature set, excavate the essential difference among REB faults and improve the accuracy of fault description. Moreover, rough set is utilized to reasonably describe the aliasing patterns and overcome the problem abnormal termination in DFV extraction process. In order to demonstrate the effectiveness of the improved method, a simulation and four fault diagnosis methods are utilized for the REB fault diagnosis. The results shown that the RS-DFV method is able to select an appropriate feature set, deeply dig the effectiveness of the features and more exactly describe the aliasing patterns. Consequently, it performs better in REB fault diagnosis than the original intelligent methods.