Detecting changes in gear surface roughness using vibration signals

The detection of gear wear from vibration signals is generally achievable once the wear has progressed to a ‘macro’ level, in which the tooth profile has changed appreciably. A typical example is the ‘double scalloped’ wear pattern – involving substantial material loss either side of the (largely unaffected) pitchline – which is diagnosable from an increase in the amplitude of the second harmonic of gearmesh frequency. Yet macro level wear is often preceded and accompanied by micro-level surface roughness changes, arising from either abrasive wear or contact fatigue pitting. These micro- and macro-level phenomena interact with one another, and so to be able to accurately predict wear rates in operating gears requires knowledge of their surface roughness state – information not easily obtainable without stopping the machine and taking detailed measurements.
This paper investigates the use of vibration signals for estimating gear tooth surface roughness. Measurements from a laboratory spur gearbox test rig are used, and the rig is fitted with gears of modified surface roughness. It is proposed that changes in surface roughness would be detectable from the nature of amplitude modulation of the random vibrations produced from asperity contacts between the teeth when they slide against one another. Such a signal – random but with cyclic modulation – is known as second-order cyclostationary, and the study finds that the degree of second-order cyclostationarity in the measured signals is strongly correlated with gear surface roughness. In comparison, the RMS and kurtosis of the vibration signal are found not to be as strongly correlated with roughness. The findings will be very important for gear prognostics, where knowledge of wear rate is critical in estimating the remaining useful life of gears.