Abstract

This work is being carried out under the framework of a thesis at the center of materials Mines -Paristech. The goal of this thesis is to realize a first digital twinning for a composite structure which is the Composite Pressure Vessel (CPV). By using the technique of data assimilation, we have a numerical model which was developed in the laboratory of Mines and which consists of a simplified stochastic FE2 model. It turns out that this model is limited, because it is too expensive in terms of computing time and moreover it is only used for fixed finite elements. We therefore used the homogenization technique to have a deterministic constitutive model, which reduces the calculation time and which liberate us from the existing mesh. With this model, we aim to use machine learning techniques to be able to learn parameters that models the generation of acoustic emission data and to be able to assimilate it in order to determine the damage of the composite pressure vessel during its service.