Compound fault diagnosis of rolling bearings based on AVMD and IMOMEDA

Abstract The intricate nature of compound fault diagnosis in rolling bearings during nonstationary operations poses a challenge. To address this, a novel technique combines adaptive variational mode decomposition (AVMD) with improved multipoint optimal minimum entropy deconvolution adjustment (IMOMEDA). The compound fault signal is isolated through AVMD, with internal parameters obtained via a new indicator termed integrated fault-impact measure index guiding the improved dung beetle optimizer. An adaptive selection method, using a weight factor, chooses the intrinsic mode function containing principal fault data. IMOMEDA whose key parameters are determined by a novel combinatorial strategy is then employed to deconvolute selected fault components, enhancing periodic fault impulses by removing complex interferences and ambient noise. The deconvoluted signal undergoes enhanced envelope spectrum processing to extract fault frequencies and identify fault types. Numerical simulations and experimental data confirm the method’s effectiveness and feasibility for compound faults diagnosis of rolling bearings, showcasing its superiority over existing techniques.