Vibration monitoring of rotating shafts using DIC and compressed sensing

Abstract

Monitoring vibrations in rotating shafts is essential for diagnosing and detecting mechanical faults. Digital image correlation (DIC), an optical full-field measurement technique, is increasingly employed in experimental mechanics. This study introduces a novel measurement approach that combines DIC with compressed sensing to monitor high-speed rotating shafts accurately. Traditionally, analyzing rotating shafts requires high-speed sampling devices, which increases experimental costs and reduces spatial resolution. To overcome these limitations, a random exposure sampling method was developed to capture speckle images on the shaft surface with a low frame-rate camera. By leveraging compressed sensing and DIC, the method reconstructs high-speed vibration signals from captured images. Step motion experiments demonstrated that the DIC system achieves a measurement accuracy of 10 µm. Experimental validation was conducted using various setups, including low-speed motors, high-speed rotating shafts, and milling machines, demonstrating the effectiveness of the approach. The measurement results of rotating shafts at 1200 rpm and 6000 rpm, compared with those obtained from laser Doppler vibrometry, demonstrated the effectiveness of the method in vibration measurement. Additionally, experiments on a milling machine showed that vibrations reconstructed using compressed sensing closely matched those measured with a high-speed camera. This measurement system shows significant potential for accurately assessing vibrations in high-speed rotating shafts, offering valuable insights for machinery monitoring and fault diagnosis.