Extraction of incipient fault features of rolling bearings based on CWSSMD and 1.5D-EDEO demodulation

Abstract

Abstract The fault feature of a rolling bearing is weak in the incipient fault stage, with severe environmental noise interference, which makes it difficult to extract the fault feature information from the vibration signal. In this paper, an adaptive method based on component-weighted symplectic singular mode decomposition and 1.5-dimensional envelope derivative energy operator (1.5D-EDEO) demodulation is proposed to extract the incipient fault features of a bearing and it does not require manual parameter setting. The method begins with the original vibration signal decomposed by symplectic singular mode decomposition to obtain multiple initial symplectic singular components (ISSCs). Then, the fault information amount of the ISSCs is measured by fault impulse sparsity (FIS) constructed by the Gini index of the square envelope which has a powerful sparsity measurement capability. After this, the ISSCs are reconstructed based on the weights obtained from the FIS to obtain the final denoised symplectic singular component (DSSC). Finally, the DSSC is demodulated by 1.5D-EDEO to further highlight the fault features of the bearing and reduce noise interference. The effectiveness of the proposed method is verified through simulation and experimental analysis. The experimental results show that the proposed method is more effective in enhancing incipient bearing fault features compared to other bearing fault diagnosis methods.