Bearing fault diagnosis based on rough set

Bearing defects were categorized as localized and distributed. For on-line bearing fault diagnosis, in this paper, the time-domain kurtosis calculation and the frequency domain wavelet analysis were used to extract the transitory features of non-stationary vibration signal produced by the bearing distributed defects. To distributed defects, bearing fault diagnosis was built on the reducing decision based on rough set. According to the information entropy and importance of attribute, adding the most important it to the reduction set, the optimization and minimum reduction set can be output by the attribute reduction algorithm, without computing the core of the attribute set. The results show that the proposed method was effective, and the method provides a promising technique of online bearings condition monitoring for practical applications.