中文摘要：

在改进形态分量分析系数阈值去噪方法的基础上，将其用于齿轮箱复合故障振动信号的分析。齿轮箱中的齿轮出现局部故障时，其振动信号中往往出现调幅调频成分；而滚动轴承出现局部故障时，其振动信号中往往出现周期性瞬态冲击成分。调幅调频成分的幅值变化相对缓慢，可看作信号中的光滑部分：而瞬态冲击成分的幅值变化较快，可看作信号中的细节部分，故可依据此形态差异实现二者的分离。采用形态分量分析方法将齿轮箱复合故障振动信号分解为包含滚动轴承局部故障信息的冲击分量、包含齿轮局部故障信息的谐振分量及残余分量，根据冲击分量和谐振分量的Hilben包络解调谱分别诊断滚动轴承和齿轮的局部故障。对齿轮箱复合故障振动信号的分析结果表明，该方法可有效分离滚动轴承与齿轮的故障特征，且效果要优于经验模态分解（Empiricalmodedecomposition，EMD）方法。

英文摘要：

Based on the improvement of coefficient threshold denoising method, the morphological component analysis（MCA） is applied to the vibration signal of a gearbox with compound fault. When a gear in a gearbox is locally damaged, the amplitude-modulation and frequency-modulation （AM-FM） component always appear in the vibration signal, and when a rolling bearing is locally damaged, the periodic transient impulse component arises in the vibration signal. The amplitude of AM-FM component varies relatively slower and thus can be considered as the smooth part of the vibration signal, while the amplitude of the impulse component changes relatively rapider and thus can be considered as the detail part, therefore, according to the morphological difference, both components can be separated. By using MCA, the vibration signal of a gearbox with compound fault can be decomposed into the impulse component containing the fault information of rolling bearing, the harmonic component containing the fault information of gear and the residual component, and then the compound fault diagnosis of rolling bearing and gear can be carried out according to the envelope spectra of the impulse component and harmonic component respectively. The results, obtained by the analysis of the vibration signals of gearbox with compound fault, indicate that the proposed method can effectively separate the fault characteristics of rolling bearing and gear, and it is superior to empirical mode decomposition（EMD） for fault feature extraction.