Rolling bearing fault diagnosis based on VMD and SABO

Abstract

To address the difficulty in extracting and identifying failure modes when rolling bearings fail during operation, this paper proposes a method that combines Variational Mode Decomposition (VMD) optimized by the Subtraction Average Based Optimizer (SABO) with Wavelet Threshold Denoising (WTD) to extract failure features. First, with the minimum envelope entropy adopted as the fitness objective, the critical parameter combination of VMD is optimized by means of the SABO algorithm. Second, the fault signal is processed for VMD decomposition with the optimized parameters acquired above; effective components are selected based on envelope entropy and kurtosis values and undergo WTD processing again, resulting (K, α) in the reconstruction of optimal signal components. Compared with other methods, this model demonstrates superior performance in rolling bearing fault diagnosis, achieving a fault recognition accuracy of 97.5578%, and thus possesses significant practical value.