Title: Image-based part programming with process parameter selection guidance for laser milling

Authors: Nikolaos Kontolatis; George-Christopher Vosniakos; Xenofon V. Gogouvitis

Addresses: School of Mechanical Engineering, National Technical University of Athens, Zografou Campus, HeroonPolytechniou 9, 15780 Zografou, Greece ' School of Mechanical Engineering, National Technical University of Athens, Zografou Campus, HeroonPolytechniou 9, 15780 Zografou, Greece ' School of Mechanical Engineering, National Technical University of Athens, Zografou Campus, HeroonPolytechniou 9, 15780 Zografou, Greece

Abstract: In this work, a software system is presented to aid tool-path and process planning in 3D laser milling, its novelty being two-fold. First, it uses images of subsequent cross-sections of the part instead of relying on 3D models and a CAM system. Zigzag tool paths are employed to clear material from appropriate areas, as necessary, and all geometry calculations are unified since they are effected in terms of pixels thereby abolishing analytical geometry. Second, appropriate metamodels (artificial neural networks) are plugged into the programming environment in order to predict machining outcome given particular values of process parameters. Any metamodel can be used after being constructed and validated from experimental, empirical or theoretical data, making for a flexible and intelligent programming environment. Demonstration of the system in CO2 laser milling of wood is followed by an assessment of profile accuracy according to the alpha-shapes method. [Received 18 September 2012; Revised 3 January 2013; Accepted 17 March 2013]

Keywords: tool path planning; pixel patterns; laser milling; image processing; part programming; process parameters; parameter selection guidance; process planning; artificial neural networks; ANNs; metamodelling; wood milling; wood machining.

DOI: 10.1504/IJMR.2013.055246

International Journal of Manufacturing Research, 2013 Vol.8 No.3, pp.313 - 335

Published online: 29 Jan 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article