Quantification and reduction of dynamically induced errors in coordinate measuring machines.

Abstract

Coordinate Measuring Machines (CMMs) are often programmed to operate at low operating speeds to reduce the negative effects of self-induced vibrations on the measurements obtained. Although these conservative operating speeds reduce structural vibration, they are detrimental to process productivity when CMMs are incorporated into a mass production environment. The trade-off between operating speed and measurement quality is well known, but up to this point a systematic approach for selecting appropriate measuring speeds has been unavailable. In this dissertation, a procedure for selecting time-optimal operating parameters used to construct motion trajectories for specified measuring paths is presented. In addition, an enhanced motion controller for use with the optimization strategy to increase CMM productivity is developed and experimentally evaluated. In Chapter 2, experimental results illustrating the effects of measurement speed on measurement quality are given. Chapter 3 is a comprehensive Analysis of Variance (ANOVA) study on the extent to which the individual factors affecting measurement time affect measurement quality. In Chapter 4, measurement quality models, based on first principles, modal analysis and experimental observations are developed for use with the optimization strategy in Chapter 5. In Chapter 6, the enhanced motion controller for CMMs is developed and analyzed. Experimental results of the optimization strategy show a significant reduction in measuring time (e.g. 25%) compared to conventional CMM speeds, without statistically significant differences in the resulting measurement quality. The optimization strategy also allows for sensitivity studies, which enable machine designers to explore the effects of changing machine characteristics on CMM productivity. The enhanced CMM motion controller design is realized through shaping controller inputs. The new shaping technique, designated Feedforward Filtering, is based on Input Command Preshaping and digital signal processing. Experimental results indicate that reductions in measurement time may be obtained for situations where high measurement quality is desired.