Stochastic modeling and identification of lubricated polymer friction dynamics.

Motivated by a strong need for an accurate model that describes transient and stochastic dynamics in friction phenomena, this dissertation is devoted to the development of a systematic modeling and identification approach to establish such a model from noisy measurements. Almost all friction models only capture steady state and deterministic dynamics of friction. In this dissertation, a novel friction model is proposed to characterize not only steady state and deterministic dynamics but also transient and stochastic dynamics. The model is derived based on physical principles that describe the energy dissipation mechanisms of friction. The structure of the model is then enriched to encompass transient and stochastic dynamics. In a short time interval, the parameters in the developed model may be considered as time-invariant. To estimate these unknown time-invariant parameters from test data, an estimation scheme based on the maximum likelihood principle is applied. In a long time interval, friction is considered to be a nonstationary process described by the developed model with time-varying parameters. To estimate these time-varying parameters, an adaptive and recursive estimation scheme based on the extended Kalman filter is applied. The models identified by both estimation schemes are finally validated by correlation tests. This work demonstrates a successful characterization of the unmodeled friction dynamics which are caused by the inaccuracy of the physically based equations, the inherent stochastic mechanisms, and the measurement noises.PhDApplied SciencesElectrical engineeringMechanical engineeringSystems scienceUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/129618/2/9542862.pd