Bearing Fault Diagnosis Using Time-Frequency Synchrosqueezing Transform*

摘要

Bearings are ubiquitous mechanical components, especially in rotating machines. Their reliability is crucial to the operation of the rotating machines. However, machine vibrations usually show periodic impulses due to bearing faults and affect the stability of machines. The common method of diagnosing bearing faults is the time-frequency analysis. However, traditional time-frequency analysis cannot discovers the dynamic characterizations of the fault signals. So, a synchrosqueezing transform using short-time Fourier transform (FSST) is employed to accurately capture the instantaneous impulse components of bearing faults. We first used a bat signal to evaluate the ability of time-frequency representation and location based on the FSST method, and further synthesized a signal consisting of several impulse components to demonstrate the accuracy. Finally, we analyzed the bearing fault data provided by the Machinery Failure Prevention Technology Society. The results prove the FSST method is a better tool that captures the instantaneous impulse signals with high accuracy, and used to diagnose the bearing faults.