关键词: 齿轮箱, 故障诊断, 信号共振稀疏分解, 最大相关峭度解卷积, 冲击特征提取

Abstract: When a rotating part of the gearbox failed, the periodic fault impulse components in the vibration signals were easily submerged by the harmonic components and the strong background noises, thus leading to the challenging difficulties in the fault feature extractions. To address this issue, a new fault diagnosis method was proposed based on RB-SSD and MCKD herein. Firstly, the low-resonance components of the signals were separated from the harmonic components and the noises using RB-SSD method, and then the MCKD method was employed to the low-resonance components to further highlight the periodic impulse components. Finally, the envelope spectrum analysis was performed to diagnose the faults. The evaluation results from simulations, experiments and engineering applications demonstrate that the proposed method may extract effectively and highlight the periodic fault impulse components from noisy vibration signals, thus providing accurate and reliable diagnosis results.

Key words: gearbox, fault diagnosis, resonance-based sparse signal decomposition（RB-SSD）, maximum correlated kurtosis deconvolution（MCKD）, impulse feature extraction