Time-Domain Sparsity-Based Bearing Fault Diagnosis Methods Using Pulse Signal-to-Noise Ratio
Chi Zhang, Shaoming Wei, Ge Dong, Yajun Zeng, Guohun Zhu, Xujuan Zhou, Feng Liu
Abstract
A fast and automated technique is crucial for bearing faults diagnosis during operation. To circumvent the intricacies of signal spectrum analysis, a diagnostic method named the pulse signal-to-noise ratio (PSNR) test is proposed by exploiting the time-domain sparsity of fault signals under a constant angular rate, which are modeled as periodic pulses with consistent duty cycle and power. The algorithm employs a statistic called pulse signal-to-noise ratio to both identify faults and determine their location. A simplified variant of the PSNR test, named pulse signal-to-noise amplitude ratio (PSNAR) test, is further proposed for near multiplication-free fast diagnosis. Data from Machinery Failure Prevention Technology (MFPT) and Case Western Reserve University (CWRU) were used to verify the algorithm.