Abstract: Aiming at the problems that the fault signal of bearing with slight wear is easily submerged by strong background noise and the fault features are weak and difficult to diagnose, a fault feature extraction method for rolling bearing with slight wear under strong background noise is proposed. The bearing vibration signal was decomposed by VMD, and the optimal intrinsic mode function was accurately selected based on the maximum kurtosis. The objective was to minimize the power spectrum entropy of the optimal intrinsic mode function, and the early termination criterion was set to realize the adaptive optimization selection of VMD parameters. The early fault signals of bearings were decomposed into several intrinsic mode functions by the VMD method with optimized parameters. The intrinsic mode function with the largest kurtosis were selected for envelope demodulation analysis, and the envelope spectrum was obtained by combining with the fast Fourier transform to achieve the extraction of fault characteristic frequency. By analyzing the measured fault signals of different types of bearings with enhanced background noise, the results show that the proposed method can effectively extract the fault characteristics of the slight wear fault signal under the interference of strong background noise, and realize the accurate diagnosis of slight wear fault of bearings, which verifies the effectiveness of the proposed method.