Non-Contact Machine Vibration Sensing and Fault Diagnosis Method Based on Event Camera

Rotating machinery plays a key role in the field of industrial manufacturing and is an indispensable equipment in the production line. Many rotating machines work in extreme environments for a long time, which makes them more prone to various failures. Now it is popular to use the vibration signal collected by the accelerometer for fault diagnosis. However, accelerometers are contact sensors, and their installations are limited in many situations. To relieve this requirement, this paper use event-based vision sensor for non-contact signal acquisition and fault diagnosis. We propose a novel event vision-based method for machine fault diagnosis. First, the event data are integrated into cumulative event frames and the time-domain signal of the vibration is extracted through a Gabor filter. Then, we use the three-layer wavelet packet decomposition method and envelope spectrum analysis on the time-domain signal to obtain the fault characteristic frequency and identify the faults. The experiments on rolling bearings validate the effectiveness of the method, indicating that event cameras have broad application prospects in the field of non-contact fault diagnosis.