MULTI-BEAD AND MULTI-LAYER PRINTING GEOMETRIC DEFECT IDENTIFICATION USING SINGLE BEAD TRAINED MODELS

Access full-text files

Date

2023

Authors

Surovi, N.A.
Soh, G.S.

Journal Title

Journal ISSN

Volume Title

Publisher

University of Texas at Austin

Abstract

In Wire Arc Additive Manufacturing (WAAM), a geometric defect is a defect that creates voids in the final printed part due to incomplete fusion between two non-uniform overlapping bead segments. Such a defect posesthe onset of a severe problem during multi-bead prints. In our earlier work, a methodology has been developed to construct machine learning (ML)-based modelsto identify geometrically defective bead segments using acoustic signals. In this paper, we investigate the performance of these single-bead segments trained defect detection model scalability for identifying voids during multi-bead prints. A comparative study of the performance of a variety of ML models is explored based on Inconel 718 material printing. The results show that the single bead segments-based defect identification model can effectively identify defective and non-defective segments in both single-layer multi-bead printing and multi-layer multibead printing.

Description

LCSH Subject Headings

Citation