Dynamic updating of the geometric model produced in additive manufacturing by wire deposit

Additive manufacturing is a process that produces parts by depositing layers on top of one another. When combined with machining and surface finishing operations, a hybrid process is obtained and increases manufacturing capabilities. Therefore, producing complex parts with multiple entities requires optimizing additive and subtractive processes. This involves adjusting and integrating various operations. Geometric defects that occur during the components' deposition can affect subsequent entity production, making real-time measurement crucial. A non-destructive measuring system of the external geometry is essential to assess the deviations without complex post-processing. A comprehensive image stereocorrelation approach addresses the challenge in this research. Therefore, an in-situ measuring system composed of a camera pair is chosen and installed in the manufacturing environment. Compared to existing ones, an improved calibration technique is proposed and adapted during this work. In addition, images of the parts produced in wire laser/arc additive manufacturing are acquired. Hence, this research suggests a multi-scale modeling approach capable of representing three levels of deviations: bead waviness, local defects, and entity scale defects. This modeling is coupled with the image correlation technique. The scale's representation is achieved by using the same pair of images while only changing the nominal geometry type: a mesh, a Bézier patch, and a superellipsoid, respectively. The three models showed interesting results that allow updating the real defective geometry in the computed aided manufacturing model. This kind of image correlation measurement in additive manufacturing hasn't been achieved in the literature. Finally, the natural texture produced in additive processes is evaluated and compared to that of ones with classical patterns. This research validated its sufficiency for image correlation and the advantage for in-situ measurement since the texturing step can be bypassed. In addition, the volume acquisition challenge is addressed. A multi-view measurement approach is implemented and validated on a case study to overcome the single-view limitations raised in this research.