Adaptive Compound Fault Diagnosis of Rolling Bearings Under Strong Background Noise: A Hybrid Framework Integrating Improved CEEMDAN With VMD Guided‐MCKD

In practical mechanical systems, the measured fault signals of rolling bearings are often submerged by strong background noise, making it difficult to extract compound fault features. To address this issue, this paper proposes a fault signal processing framework that integrates Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN), Variational Mode Decomposition (VMD) guided‐Maximum Correlated Kurtosis Deconvolution (MCKD). First, an efficient decomposition optimization strategy and an adaptive mode selection scheme are developed to achieve robust preliminary denoising of raw fault signals using ICEEMDAN. Then, based on the Grey Wolf Optimizer (GWO), a novel Weighted Envelope Harmonic Energy Ratio (WEHER) is proposed as a cost function to adaptively optimize the key parameters of VMD, automatically extracting the mode that carries the richest fault information. This process enables effective extraction of compound fault features even under strong noise and provides accurate estimates of fault characteristic frequencies. Finally, the estimated frequencies are subsequently utilized as prior guidance for the MCKD algorithm, with the Fault Energy Ratio (FER) as the cost function, to adaptively optimize the parameters of MCKD based on GWO, thereby further enhancing the weak fault‐induced impulses with high precision. Simulation and experimental results demonstrate that the proposed method can effectively extract compound fault features of rolling bearings under strong background noise, with superior accuracy and robustness over conventional methods, showing significant potential in practical diagnosis.