In view of the fault feature extraction of rolling bearings that are susceptible to random noise

this paper proposes a fault feature extraction algorithm based on Infogram and improved MNAD (Minimum noise amplitude deconvolution). Firstly

Infogram is used to obtain the optimal frequency band and bandwidth for band-pass filtering

thereby reducing the influence of noise components. Secondly

the multipoint kurtosis spectrum is calculated on the preprocessed signal

the square envelope Gini coefficient (SEGI) is applied to the filter length selection of MNAD

and the optimal filter length for MNAD is selected adaptively. Finally

the MNAD is optimized based on the multipoint kurtosis spectrum and the optimal filter length

and the coordination of rolling bearings is combined with the envelope spectrum to realize fault diagnosis. The analysis of the simulation signals and experimental data proves the effectiveness of the proposed method. At the same time

it is applied to the engineering measured signal to verify its deconvolution ability in the actual working conditions.