中文摘要：

为了从复杂的轴承振动信号中提取微弱的故障信息,提出了一种基于局部均值分解（local mean decomposition,LMD）和奇异值差分谱的轴承故障诊断方法。首先通过LMD将非平稳的原始轴承故障信号分解为若干个PF（product function）分量,由于背景噪声的影响,难以从PF分量准确得到故障频率,对PF分量进行Hankel矩阵重构和奇异值分解,相应的得到奇异值差分谱,根据奇异值差分谱理论对某个PF分量进行消噪和重构,然后再求重构后PF分量的包络谱,便能准确地得到故障频率。仿真分析和滚动轴承内圈故障实例很好地验证了提出的改进方法的有效性。

英文摘要：

In order to extract faint fault information from the complicated vibration singals of rolling bearing,a novel bearing fault diagnosis method is presented based on the principles of local mean decomposition（ LMD） and difference spectrum of singular value. In this method,the original nonstationary fault signals of rolling bearing were decomposed by LMD and a group of product functions（ PFs） were obtained; However,it is difficult to extract fault frequencies due to strong background noise. To identify the fault pattern,a Hankel matrix of the PFs was constructed and decomposed with singular value decomposition（ SVD）. Accordingly,difference spectrum of singular values was also obtained. On the basis of difference spectrum theory,the filtered and reconstructed signal of the PF component was analyzed by using envelope spectrum and the acurate fault frequency can be obtained. The effectiveness of the proposed method was demonstrated with the simulated data and the actual signals measured in the inner race of a fault rolling bearing,and the present method had a good prospect for its application in rolling bearing fault diagnosis.