Traitement des données volumétriques par corrélation virtuelle : Application à l'identification de défauts géométriques en fabrication additive

Additive manufacturing is a relatively new process compared with traditional manufacturing processes. The various existing additive processes involve complex physical phenomena, combined with innovative machinery and a variety of materials. As a result, additive manufacturing calls into question the tools developed and mastered for other, more traditional processes. In particular, in the field of shape acquisition, conventional measuring instruments are unsuitable for determining the geometry of parts with external and internal shapes, often affected by a panoply of defects of their own, with multiple origins. In this work, lattice structures are studied in particular. These consist of the three-dimensional repetition of an elementary pattern in three directions in space. X-ray measurement overcomes these accessibility problems and provides volumetric information about the part being measured, containing information about the geometric defect. However, volumetric data is difficult to manipulate before extracting the information of interest. It requires a succession of steps that are highly dependent on the user and their experience. For this reason, this work proposes a methodology for extracting information about the measured defect directly from the volumetric data, without any steps requiring user intervention, or thresholding choices that erase part of the information of interest. The proposed methodology is called virtual volume correlation and is based on an a posteriori representation of the geometric defect encountered in strut-based lattice structures.