Entropy guided correlated ACMD and Enhanced CYCBD for bearing fault diagnosis

The failure characteristics of bearings can be difficult to detect. This is due to the influence of strong background noise. To accurately identify the fault characteristics of bearings, a method for extracting features by fusing Adaptive Chirp Mode Decomposition (IACMD) in view of entropy value and correlation coefficient regrouping scheme with the Second-order Cyclostationarity Blind Deconvolution (ECYCBD) based on the sparrow search algorithm and the summation of weighted harmonic (WHS) is proposed in the paper. This method integrates the decomposition capabilities of ACMD for multi-component non-smooth signals and the signal amplification properties of CYCBD. Initially, the signal undergoes decomposition into a string of modal signals using ACMD, followed by the reorganization of these modal signals through a proposed scheme based on entropy value and correlation coefficient. Subsequently, the sparrow search algorithm is used to optimize the length of CYCBD filter. By employing the WHS as an optimization index, all frequencies within the signal spectral range are examined as potential candidates, with the frequency yielding the highest WHS being considered the optimal cyclic frequency. Finally, the reconstructed components are processed through the ECYCBD method to filter the signal,and the square envelope analysis is used to extract features.