For the fault diagnosis of rotating machinery, the demodulation algorithm of the monitoring signals plays a key role in fault feature extraction. Especially for weak fault features extraction, existing single narrow band demodulation methods have worse performance under low signal to noise ratio condition. According to the mechanism of rotating machinery, both narrow and broad frequency band modulated signals exist simultaneously. Therefore, weak fault features can be obtained through demodulation of multiple narrow frequency bands rather than only one resonance narrow band. In this study, a novel weak feature extraction method is proposed based on used as a good filtermultiple frequency bands demodulation. The superiority of the proposed method is corroborated by simulation analysis and applications of a centrifugal pump and a propeller. By simulation analysis, the proposed multiple frequency bands demodulation (MFBD) method has better demodulation performance than Fast Kurtogram, Autogram and Fast-SC for weak modulation features. The applications results suggested that the proposed MFBD provided a clearer characteristic frequency identification than Fast Kurtogram, Autogram and Fast-SC, especially in weak modulation condition. Therefore, the proposed MFBD method provides a reliable basis for weak fault signal extraction, which shows good engineering significance for fault diagnosis of rotating machinery and passive detection of propeller.

对于旋转机械的故障诊断，监测信号的解调算法在故障特征提取中起着关键作用。特别是对于弱故障特征提取，现有的单一窄带解调方法在低信噪比条件下性能较差。根据旋转机械的机理，窄带调制信号和宽带调制信号同时存在。因此，可以通过解调多个窄频带而不是仅一个谐振窄带来获得弱故障特征。在这项研究中，提出了一种新的弱特征提取方法，该方法基于用作良好滤波器的多频段解调。离心泵和螺旋桨的仿真分析和应用证实了该方法的优越性。通过仿真分析，所提出的多频带解调（MFBD）方法对于弱调制特征具有比Fast Kurtogram、Autogram和Fast-SC更好的解调性能。应用结果表明，所提出的MFBD比Fast Kurtogram、Autogram和Fast-SC提供了更清晰的特征频率识别，尤其是在弱调制条件下。因此，所提出的MFBD方法为微弱故障信号的提取提供了可靠的依据，对旋转机械故障诊断和螺旋桨被动检测具有良好的工程意义。应用结果表明，所提出的MFBD比Fast Kurtogram、Autogram和Fast-SC提供了更清晰的特征频率识别，尤其是在弱调制条件下。因此，所提出的MFBD方法为微弱故障信号的提取提供了可靠的依据，对旋转机械故障诊断和螺旋桨被动检测具有良好的工程意义。应用结果表明，所提出的MFBD比Fast Kurtogram、Autogram和Fast-SC提供了更清晰的特征频率识别，尤其是在弱调制条件下。因此，所提出的MFBD方法为微弱故障信号的提取提供了可靠的依据，对旋转机械故障诊断和螺旋桨被动检测具有良好的工程意义。