中文摘要：

滚道损伤会在轴承振动中引起调幅现象,但当损伤较轻微或者噪声干扰较严重时,这种特征难以显现出来。提出利用奇异值分解（Singular value decomposition,SVD）来提取这种调幅特征。在Hankel矩阵方式下,SVD可将信号分解为一系列分量信号的简单线性叠加。利用这一特性,对轴承振动信号构造Hankel矩阵并进行SVD处理,通过选择合适的分量进行简单相加,从一个复杂的信号中准确提取了清晰的调幅特征信息,据此确定了载波和调幅波频率,并结合轴承参数对滚道损伤状况做出了准确的判断,这种特征提取效果远优于带通滤波器的处理结果。

英文摘要：

Amplitude modulation will be caused in bearing vibration if roll track is injured,but this modulation phenomenon is hidden in the raw vibration signal when injury is slight or there is noise interference,and singular value decomposition（SVD） is proposed to extract this modulation feature.By virtue of the Hankel matrix,SVD can decompose a signal into the linear superposition of a series of component signals.This characteristic being utilized,for a complicated vibration signal of bearing,a Hankel matrix is created using this signal and is processed by SVD,then the front several component signals are selected to be added together,and the result shows that the clear amplitude modulation signal is got,frequencies of the carrier and the modulated wave are both precisely obtained,and the reliable diagnosis conclusion for roll track injury state is also made according to this feature signal.This feature extraction effect of SVD is much better than that of band pass filter.