Rolling bearing is the wearing part

and from slight fault to damage expand quickly.Therefore

found rolling bearing initial slight fault is great significance.The initial slight fault characteristics of bearing are diagnosed based on combining the wavelet packet entropy and EMD.Signal-noise separation is achieved by wavelet packet

the de-noising effect of the wavelet packet method is obvious.Extracting The signal components decomposed and de-noised by EMD is extracted with cross-correlation

kurtosis criterion

the blindness of IMF components selection is avoided.Through the simulation signal analysis and case analysis

the bearing fault can be accurately detected

it is shows that the method is effective.