Vibration based gear wear monitoring and prediction

Abstract

Gear wear is an inevitable phenomenon during gear service life. Its propagation would impair the durability of gear tooth and reduce the remaining useful life of gear transmission system. Therefore, monitoring and predicting gear wear progression can bring significant benefits to industrial practice. Vibration analysis responds immediately to changes in the machine state (health and operating condition) and can therefore be used for gear monitoring. However, vibration-based techniques for gear wear monitoring are rather rare, even though techniques have been well established for detection and diagnosis of common gear faults such as gear tooth root cracks and tooth breakage. Therefore, in this research, a vibration-based integrated system is developed for gear wear monitoring and prediction. The developments were carried out in two stages: (i) wear mechanism identification using measured vibrations, and (ii) wear propagation monitoring and prediction using the integration of models, measurements and model updating approaches.

In the first stage, the correlation between surface features and vibration characteristics is investigated. Then, use of cyclostationary properties of vibrations, a vibration-based online gear wear mechanism identification methodology is developed. Moreover, the evolution of fatigue pitting and abrasive wear (micro-level) are tracked using an indicator of second-order cyclostationarity of vibrations in specific spectral bands.

In the second stage, a digital-twin system is developed by the integration of (i) a dynamic model to simulate the dynamic responses of gear system; (ii) two tribological (wear) models for estimation of wear depth and pitting density, and (iii) model updating through comparing simulation and measured vibrations. The integration of dynamic model and tribological models allow a knowledge-based wear prediction of the gear profile change (determined by the wear depth) and pitting density. With the regularly model updating using measured vibrations, the wear process can be well monitored, and the best possible prediction of remaining useful life can be achieved.

The above developments provide effective and efficient tools for monitoring and prediction of gear wear, in particular, the profile change and pitting density, which is critical for making appropriate maintenance decisions to maximise the useful life of gears and to avoid catastrophic failures and unexpected economic losses.