Machine tool accuracy enhancement by inverse kinematic analysis and real time error compensation

Real time error compensation (RTEC) has been one of the most successful techniques to reduce geometric errors as well as thermally induced errors of machine tools. However, long pre-process calibration time and lack of a generalized error kinematics modeling and compensation technique limits the application of this technology.; In this research, a generalized error kinematics modeling technique was developed. This technique can be used to derive error synthesis models for machines with both prismatic and rotary joints. The machine volumetric (or planar) error was described by an error matrix which includes position and orientation errors rather than an error vector which only includes position errors. A generalized error compensation strategy was also developed. An inverse kinematics technique was applied to obtaining compensation signals for machine joints such that machine errors were minimized.; A fast error identification scheme for both geometric errors and thermally induced errors was developed. The inverse kinematics technique was employed to identify machine errors based on measurements made by alternative measurement devices (telescoping ball bar, laser tracker, etc.). These devices can acquire data in a relatively short time compared to a laser interferometer system. An error synthesis model which was derived automatically by the generalized error kinematics modeling method was used to relate the measurements to individual error components. Then a least squares estimation was applied to estimating the coefficients of error component models.; The error identification method was applied to a 3-axis machining center and a 3-axis optical coordinate measuring machine (OCMM). The calibration on the machining center was performed using a telescoping ball bar in three hours for the geometric errors and one day for the thermal errors. The maximum error of the machine was reduced from 320 {dollar}mu{dollar}m to 30 {dollar}mu{dollar}m by the RTEC based on the identified error models. For the OCMM, the geometric errors were identified using a laser tracker in two hours, and the error prediction accuracy was about 100 {dollar}mu{dollar}m while the original machine accuracy was about 2500 {dollar}mu{dollar}m.