Time–frequency analysis of tribological systems—part I: implementation and interpretation

Abstract

A novel technique adapting the time–frequency analysis has been utilized to characterize stationary and non-stationary signals from tribological interactions. This representation displays time, frequency, and signal magnitude to decipher signals emanating from such interactions. Short-time Fourier transform, Wigner, Coi–Williams, and Zhao–Atlas–Marks distributions are suited to represent stationary and non-stationary signals. Some of the most complex tribological phenomena involve head–disk interactions in magnetic recording systems. Examples drawn from practical head–disk interface tests are analyzed by using the fast Fourier transform algorithm to illustrate the dynamic features of various distributions. Time–frequency representation of output spectrums of laser doppler vibrometer (LDV), strain gage sensor, and acoustic emission (AE) sensor obtained from head–disk experiments giving evidence of stationary and non-stationary behavior are investigated.