Machine learning driven instance segmentation providing new porosity insights into wire arc directed energy deposited Ti-22V-4Al

• DOI: 10.1016/j.addma.2024.104323
• Publication Date: 2024-07-31T03:16:55Z
• Authors: Yuan Wang; Chi-Ho Ng; Michael Bermingham; Matthew Dargusch
• Keywords: Sphericity

Non-destructive x-ray methods such as micro-computed tomography (micro-CT) are useful for investigating porosity defects in additively manufactured products. Understanding the knowledge accessible to micro-CT technologies relies on quantifying spatial and morphological characteristics of instances. So far, machine learning-based semantic segmentation techniques threshold-based image analysis have been employed this purpose. However, reliability these is compromised by their limitations accuracy delineating connected pores. This work proposed a investigation strategy involving deep instance process slices followed 3D reconstruction reliably examine data porosities, cased studied wire arc directed energy deposited Ti-22 V-4Al alloy. Systematic captured new insights pore terms formation tendency, relationship between size sphericity, evolution, discovery pore-free zones. The workflows transferrable other components. may also guide future advancements elimination manipulation fabricate mechanically robust complex structured