A novel CYCBD β deconvolution for weak bearing fault detection enhanced by optimized successive jump mode decomposition

Abstract

Bearings are critical components in rotating machinery, making fault detection essential for operational safety and reliability. However, weak bearing faults are difficult to detect due to strong noise interference and the masking effect of background vibrations. To address these challenges, this paper proposes a novel weak fault detection method by combining optimized successive jump mode decomposition (OSJMD) with a blind deconvolution technique based on generalized Gaussian cyclostationarity (CYCBD_β). CYCBD_β offers superior robustness against both Gaussian and non-Gaussian noise compared to existing methods. SJMD effectively separates noise from raw signals and extracts impulse-like features. To enhance SJMD’s performance, a synergistic swarm optimization (SSO) algorithm is used to optimize the key parameter—mode compactness α. By integrating OSJMD and CYCBD_β, the proposed method significantly improves the detection of weak fault signals. Simulation and experimental results demonstrate its effectiveness and superiority over conventional approaches.