Gear condition monitoring by augmenting measured transmission error data for gear damage and propagation estimation

Structural changes and damage to gears can lead to critical failure of gear transmission systems. However, current condition monitoring approaches often lack precision due to the limited availability of labelled run-to-failure data. Therefore, we provide an approach to augment measured transmission error data with simulated data. The simulation is based on tactilely measured gear flanks with (micro)pitting of varying severity. We automatically identify (micro)pitting with respect to the gear flank and subsequently simulate geometrical and temporal damage growth. This allows us to compute the temporal change in (micro)pitting expansion, as well as the resulting transmission error. By comparing this simulated with the measured transmission error of the running gear transmission, an estimate of the current degree of gear damage is possible. The presented approach offers to combine state-of-the-art damage propagation models with a dataset of measured gear flanks and data augmentation to determine the health condition of gear transmissions.