Abstract: Aiming at the problem that the compound fault feature of the roller bearings of the wind turbine is weak and the different faults are coupled with each other in the background of strong noise,the fault feature is hard to be extracted.a compound fault separation method of rolling bearing based on multipoint kurtosis multipoint optimal minimum entropy deconvolution adjusted（MK-MOMEDA）and Teager energy operator is proposed.Firstly,deconvolution multipoint kurtosis spectrum analysis is performed on the compound fault signal to determine the period of the fault impact signal.The feature extraction interval including the fault period is respectively set according to the determined fault period,and then the fault signal is deconvolved and the fault characteristics is separated,the impact component is enhanced by the Teager energy operator after deconvolution,and finally the enhanced signal is made FFT,and the compound fault feature can be effectively identified by analyzing the dominant fault characteristic frequency in the spectrum.The experimental platform simulates the compound fault condition of the rolling bearing and the wind turbine bearing compound fault to verify the proposed algorithm.The results show that the proposed method can effectively separate various fault features from the compound fault and successfully diagnosis the fault type.