Encogram: An autonomous weak transient fault enhancement strategy and its application in bearing fault diagnosis

Extracting weak repetitive transients is crucial for rolling bearing fault diagnosis. Most existing periodic fault feature extraction techniques are prone to be affected by the interference of transmission paths, background noise, and incidental shocks. One of the primary reasons is that a constant spectrum segmentation strategy may result in a loss of significant fault information. Another reason is that many indicators do not simultaneously consider impulsiveness and cyclostationarity of fault transient impulses. Therefore, an autonomous weak transient fault enhancement strategy called Encogram is proposed in this paper, which can adaptively segment frequency bands based on spectrum distribution characteristics and immunize an influence of inevitable incidental shocks and strong background noise. Compared to Kurtogram, Autogram, and other state-of-the-art signal processing methods, the proposed Encogram can obtain a frequency band with rich fault information and less noise interference.