2025 International Conference on Equipment Intelligent Operation and Maintenance (ICEIOM2025)

37 / 2025-05-12 16:09:21

Feature Extraction of Machine Faults Based on Envelope Matrixes of Source Signals

blind source separation; envelope demodulation; singular value decomposition; feature extraction

In mechanical signal processing, vibration signals are mostly amplitude-modulated or frequency-modulated signals. Moreover sensor signals are always the aggregates or mixtures of independent vibration sources generated by different mechanical components. The BSS algorithms can recover or separate independent vibration sources from sensor signals. The envelope demodulation analysis is known as an effective approach to process modulated signals. Thus a feature extraction method based on envelope matrixes of source signals is presented to be applied to performing feature extraction and fault diagnosis of rotating machines. The experimental results of hydraulic gear pump indicate that feature information extracted from envelope matrixes of source signals can accurately represent the machine states.