Cooling fan bearing diagnosis based on AR& MED

To monitor the initial failure of cooling fan´s rolling bearing, this paper reviews the theory of autoregressive (AR) model and the minimum entropy deconvolution (MED) filtering technique. The AR method can remove the deterministic components of the original signal, and the MED filter could reduce the effect of the transmission path. These two filtering techniques were combined in this paper to pre-process the rolling bearing´s vibration signal, and then the envelope spectrum of the residual signal was analyzed. The method leads to efficient filtering result.