Friction cancellation in flexible systems using extended kalman-bucy filtering

Abstract

Positioning systems that are subject to nonlinear friction are often accompanied by substantial flexibility and thus require friction compensation in the presence of flexibility. This paper develops and demonstrates extended KalmanBucy Filter (EKBF) based friction estimation for friction compensation in such systems. Building upon successful application of friction estimation using an EKBF for rigid systems, a friction observer is developed for a positioning system consisting of a gear driven inertia with flexibility in gear tooth interactions. The EKBF-based observer uses motion memurements and the system dynamic model to estimate an unknown friction torque. Two configurations of the friction observer are described and evaluated. The first configuration estimates the friction torque directly and uses the resulting estimate for friction cancellation, while the second configuration estimates a single parameter of the simple Dah1 friction model. Each method is demonstrated by numerical simulation of sinusoidal position tracking experiments, in which the underlying friction characteristics are completely unknown to the friction observer. Comparison of EKBF-based friction estimation and compensation to an adaptive model-based method demonstrates that EKBFbased compensators reduce tracking error significantly and introduce less complexity than model-based methods.