机械设备发生故障时, 故障特征的提取很重要。对于多通道的设备故障振动信号, 应用非平稳信号的盲源分离算法, 可以有效地提取各自独立的非平稳振动源,从而可以准确地进行机械故障诊断。针对不同时频分布的非平稳盲源分离算法, 比较了它们的分离效果。以转子的复合故障为例, 验证了该算法在故障诊断中可行性。-Machinery and equipment failure, the fault feature extraction is very important. Failure vibration signal for multi-channel devices, the application of non-stationary signals blind source separation algorithm can effectively extract the independent non-stationary vibration source, which can accurately mechanical fault diagnosis. Not the same frequency distribution of non-stationary blind source separation algorithm to compare the effect of the separation. To the composite rotor failure, for example, to verify the feasibility of the algorithm in fault diagnosis.