During the operation cycle of geared transmissions, remaining useful lifetime predictions are key information for maintenance and future design decisions. These predictions are usually obtained by combining modelling efforts with limited sensor data. This contribution proposes an estimation method that aims to be robust to the operating conditions and the statistical properties of the introduced damage. As damage on the tooth surface causes a reduction in the gear pair mesh stiffness due to the changed contact conditions, this time-varying mesh stiffness is proposed as health indicator. The mesh stiffness is parameterized in a piecewise interpolation scheme to guarantee the local detectability of the involved parameters. The parameters are estimated concurrently with the states using an augmented extended Kalman filter. The method is applied on a single helical gear pair of an industrial gearbox, combining a lumpedparameter contact model with angular position data of both gear shafts. For the validation, virtual measurement data are generated using a gear contact model with pitting defects. The estimation results show a proof of concept and highlight the potential of the method for more complex cases.

Reliable system-level simulation tools capable of modelling and predicting the physical interactions between components are of special interest in the analysis of mecha(tro)nic drivetrains. In such drivetrains the flexibilities of shafts and bearings alter the gear alignment conditions, thereby requiring gear contact models that are capable of capturing the associated misalignment effects. In this contribution, an efficient yet accurate model for misaligned helical gear contact analysis is derived by lumping a distributed-parameter model that was recently developed by the authors and validated numerically by comparison with finite element simulations. The transformation into a lumped-parameter model relies on a computationally efficient linearization approach that can be generalized to other gear contact models. The developed model is numerically validated on the component-level by comparison with finite element simulations of a helical gear pair considering various misalignments. The twisting rotation is shown to be the most influential misalignment, significantly altering the transmission error and thereby dynamic performance of the gear pair, and is well-captured by the developed lumped-parameter model. In order to demonstrate the model’s capabilities and computational efficiency in a system-level simulation, the model is implemented in a port-based multi-physical simulation environment, where it is used to simulate a two-stage helical automotive transaxle.