A sub-Nyquist sampling and data compression approach for efficient ultrasonic measurement of bearing lubrication

Ultrasonic measurement of bearing lubrication is crucial for monitoring the condition of rolling bearings. However, conventional methods that sample high-frequency ultrasound signals above the Nyquist rate require extremely high hardware sampling frequencies, leading to large data volumes and presenting significant challenges for data acquisition, storage, and transmission. Motivated by the inherent sparsity of ultrasound signals, particularly in bearing lubrication measurements, this paper proposes a novel sub-Nyquist sampling and data compression method for ultrasonic measurement of bearing lubrication. The method combines a sparse representation of ultrasound signals, achieved through an ultrasound-customized dictionary (UCD), with sub-Nyquist sampling via random demodulation. By exploiting the inherent sparsity of the ultrasound signals, the proposed approach significantly reduces both the sampling rate and data volume, while maintaining high reconstruction accuracy. Experimental results demonstrate the effectiveness of the method. With a sampling frequency of 0.5 Nyquist rate, the mean error of the reconstructed reflection coefficient curve is only 0.75%, and the acquired ultrasound data are compressed from 640 MB/s to 34.85 MB/s. The reduced sampling rate and data volume make the proposed method particularly suitable for real-time, large-scale industrial applications, providing a promising low-cost solution for ultrasonic monitoring of bearing lubrication.