Application of an Unscented Kalman Filter for Modeling Multiple Types of Machine Tool Errors

This paper introduces the application of the Unscented Kalman Filter (UKF) for the support of different error models for machine tools. The UKF is used to minimize measuring and modeling errors for geometric and thermal errors of machine tools. Error models are introduced and transformed into a formulation for the UKF. Geometric and thermal error measurements in three- and five-axis machine tools are presented. The modeling and potential correction results with and without the UKF are analyzed and compared. It is observed that the Unscented Kalman Filter is able to reduce modeling errors due to non-linearity and measurement noise.