Abstract: In the early fault of rolling bearing,the impact component in the signal is seriously disturbed by the noise,it is difficult to extract the periodic feature of the fault signal.Aiming at solving this problem,incipient fault diagnosis method for rolling bearings was proposed based on Local Mean Decomposition(LMD) and Maximum Correlated Kurtosis Deconvolution(MCKD).Firstly of all,the LMD was adopted to collect original fault signals,the Product Function(PF) component of the original signal correlation coefficient is selected to reconstruct the signal.Then,the reconstructed signal is processed by the MCKD,prominenting periodicimpluse component.Finally,the processed signal is analyzed by Hilbert envelope.The fault characteristic frequency can be extracted accurately from the envelope spectrum.Through the simulation and the analysis of the experimental data of inner fault,the effectiviness of the proposed method was berified.

Key words: bearing, early fault, LMD, MCKD