When the flexible thinwalled bearing works,it is difficult to extract the fault frequency from the spectrum of vibration signal because of the influence of the impact component alternately generated by the long and short axes and the influence of background noise.Aimed at this problem,a fault feature extraction method for flexible thinwalled bearings combined with singular value decomposition (SVD) and multipoint optimal adjustment minimum entropy deconvolution (MOMEDA) was proposed.In this method,the original signal is denoised by SVD algorithm to obtain the reconstructed signal,the reconstructed signal is enhanced by MOMEDA to highlight the periodic fault pulse,and the processed signal is analyzed by spectrum to extract the corresponding fault frequency.By comparing the dominant frequency in the spectrum with the fault characteristic frequency of the flexible thinwalled bearing,the fault location can be judged and the fault diagnosis of the bearing can be realized.The test data analysis results show that the method can effectively extract the fault frequency of the inner and outer rings of the bearing.