中文摘要：

压缩感知作为一种新型压缩采样方法，利用信号稀疏特性以远低于奈奎斯特采样定理的采样速率压缩采集信号，减小数据采集、传输、存储的硬件压力。基于压缩感知框架下压缩采集的信号，提出了一种滚动轴承故障诊断新方法。该方法选择部分hadamard矩阵作为测量矩阵，将峭度因子、方差、波形因子作为敏感特征参量，不重构压缩测量量，直接利用压缩采集信息，提取敏感特征，然后通过PS0-SVM算法进行模式识别从而实现故障诊断。研究结果表明，在一定压缩比范围内，利用该方法能够在降低平均采样速率的同时用更少的数据量表现故障特征，实现滚动轴承故障诊断。

英文摘要：

As a new sampling method, compressed sensing samples with the signal sparse features were presented, which was far below the Nyquist sampling theorem. It might reduce generous re- quirements of data acquisition, transmission and storage hardware. Aiming at the signals from the compression perception within the framework, this paper proposed a new method for rolling bearing fault diagnoses. In this method, part of hadamard matrix was chosen as a measurement matrix, and kurtosis factor, variance and waveform factor as a sensitive parameters. So there was no need to re- build compression measurement and the gathering informations were utilized to extract sensitive char- acteristics directly, and then the PSO-SVM algorithm was used for pattern recognition so as to realize fault diagnoses. It is shown that within a certain range compression ratio, the method may use less a- mount of data of fault characteristics for rolling bearing fault diagnoses.