Contributions au développement d'algorithmes robustes pour l'identification des défauts géométriques de forme à partir d'un nuage de points

Abstract

The improvement of the analyze resulting from the exploitation of the digital model requires the use of realistic models which reflect the variations induced by the manufacturing processes. Realistic component modeling must consider these variations. More specifically, we are aiming for the present project, the modeling of geometrical shape tolerances, the simulation of assemblies of parts with defects of form and the evolution of the effect of the algorithmic error on the estimation of the dimension of surfaces with form defects. The developed models present an analysis tool to assist the engineering of the methods during the creation of the ranges manufacturing and inspection. The first step is to model form defects for various primitive geometries considering infinite rigidity. At this stage, we develop algorithms that automatically generate realistic assemblies. This realistic modeling allows a tolerance analysis taking into account the evolution of the contacts between the components during the operation of the product. This approach has provided an assembly result closer to the actual assembly of the mechanical system. As a second step, a study on the influence of form defects on the identification of dimensional and geometrical requirements simulated from a point cloud measured. Specifically, we examine the use of new size tolerance specification modifiers, as defined in ISO 14405-1: 2016, and investigate how system measurement strategies and algorithms affect the evaluation of these new specifications. In addition, computer-assisted inspection systems are well used by manufacturers to check the conformity of the parts they produce. However, the results of these inspections may vary depending on the software and the parameters used. This is what researchers call "algorithmic error". We are conducting a study to examine the reproducibility of computer-assisted inspection software in the case of assessing form defects. Finally, a new approach is proposed for estimating the size of geometric elements affected by a form defect using the robust principal component analysis and the oriented bounding box. This new approach has been validated by a numerical simulation and on real studies.

Date	19 Nov 2019
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Antoine Tahan (Supervisor) & Mohamed Ali Mahjoub (Co-supervisor)