Quasistatic Error Modeling and Model Testing for a 5-Axis Machine

Hua Wei Ko, Patrick Bazzoli, J. Adam Nisbett, Le Ma, Douglas Bristow, Robert G. Landers, Yujie Chen, Shiv G. Kapoor, Placid M. Ferreira

Research output: Contribution to journalArticlepeer-review

Abstract

This paper presents an approach to modeling the quasistatic errors of a 5-axis machine tool with one redundant axis. By introducing errors to the ideal joints and shape transforms of the kinematics of the machine, an error model is obtained. First order error characteristics are used to parameterize the introduced errors. It is found that of the 52 introduced error parameters, only 32 have a linearly independent effect on the volumetric errors observed in the machine's workspace. To identify these error parameters, the volumetric error components at 290 randomly chosen points are measured with a laser tracker. The unknown parameters are obtained by least-squares estimation, and the resulting model able to reduce average magnitude of the volumetric error vectors at these points by an average of 90% of their original values. Further, the identified model was used to predict the errors observed in two independent test point sets (each set consisting of 48 points). A 75% reduction in the average magnitude of the error vectors was observed. A large fraction of the residual errors was found to be attributable to the thermal drift of the machine during the experiments where were not conducted in a thermally controlled environment and the positioning repeatability of the machine.

Original languageEnglish (US)
Pages (from-to)443-455
Number of pages13
JournalProcedia Manufacturing
Volume10
DOIs
StatePublished - 2017

Keywords

  • error compensation
  • machine-tool accuracy
  • machine-tools
  • quasistatic machine-tool errors

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering
  • Artificial Intelligence

Fingerprint

Dive into the research topics of 'Quasistatic Error Modeling and Model Testing for a 5-Axis Machine'. Together they form a unique fingerprint.

Cite this