A High-Performance Bearing Fault Diagnosis Enhancement Method in High Noise Environments Based on PSO-Optimized VMD-MCKD

Addressing the challenge of extracting faint fault features of rolling bearings in environments with substantial background noise, this study introduces a novel diagnostic method for bearing faults. This method harnesses a Particle Swarm Optimization algorithm to sequentially fine-tune two parameters of Variational Mode Decomposition (VMD), [α, K], and three parameters of Maximum Correlation Kurtosis Deconvolution (MCKD), [L, T, M]. Utilizing the bearing fault data from Case Western Reserve University, this approach has been substantiated to significantly mitigate noise interference and accentuate faint fault features in high-noise settings. The experimental outcomes affirm that this method effectively extracts subtle fault characteristics, thereby enhancing the precision and reliability of fault diagnosis.