In order to solve the problems that the fault feature of rolling bearing is difficult to extract, the signal analysis method of variational mode decomposition(VMD) combined with independent component analysis(ICA) was proposed in this paper. At first, VMD was used to decompose the multi-component signal into a number of quasi-orthogonal intrinsic mode functions. To a great extent, the problems existed in the LMD algorithm, such as mode mixing and end effect were eliminated with help of the proposed method. Then the IMFs were reconstructed as the input matrices of ICA based on kurtosis criterion. After that, the source signal and noise signal can be separated with the help of FastICA algorithm, in addition, the fault features can also be extracted. In order to verify the effectiveness of the proposed method, the simulation experiments and practical engineering experiments had been carried out. The effect was compared with the LMD-ICA algorithm for rolling bearing inner fault signals. The results demonstrate that the proposed method can not only solve the problems such as losing fault information and leaving noise due to the mode mixing in the process of denoising, but also extract the fault frequency accurately.