Generative adversarial networks for tolerance analysis. Issue 1 (2022)

• Record Type: Journal Article
• Authors: Schleich, Benjamin; Qie, Yifan; Wartzack, Sandro; Anwer, Nabil
• Is Part Of: CIRP annals. Volume 71:Issue 1(2022)
• Page Start: 133
• Page End: 136
• Publication Date: 2022
• Subjects: Design -- Tolerancing -- Machine learning; Production engineering -- Research -- Periodicals; 670.5
• DOI: 10.1016/j.cirp.2022.03.021
• Languages: English
• ISSNs: 0007-8506

Abstract

Abstract: Many activities in design and manufacturing rely on realistic product representations considering geometrical deviations to assess their effects on the product function and quality. Though several approaches for tolerance analysis have been developed, they imply several shortcomings, such as the lack of form deviations consideration and the high manual modelling effort. In this paper, a novel shape-agnostic approach supported by generative adversarial networks is developed for the automated generation of part representatives with geometrical deviations. A workflow for generating these variational part representatives is highlighted and tolerance analysis case studies demonstrate the effectiveness of the proposed approach.