The characteristics of the rolling bearing fault signal are easily drowned by strong noise,and it is difficult to extract the impact component in the signal.To solve this problem,a multipoint optimal minimum entropy deconvolution adjustment (MOMEDA) optimized ACCUGRAM algorithm was proposed and applied to the fault diagnosis of rolling bearings.The MED algorithm was used to filter the original signal to highlight the effective cyclic impact components in the signal,and the classification accuracy of the band selection in the ACCUGRAM algorithm optimized by MOMEDA was improved.Then the optimal bandwidth and center frequency could be selected.Finally the fault characteristic frequency was extracted from the frequency band containing the most information.The results of simulation and test show that this method can be used to effectively extract the periodic shock characteristics in the signal,and has certain practicability.