Representation of FEM-model by neural networks for optimisation of SLS-processes

• Unité de recherche: COST
• Numéro de projet: C00.0071
• Titre du projet anglais: Representation of FEM-model by neural networks for optimisation of SLS-processes

Mots-clé (Anglais)

Neural Network; finite element model; non-linear process; process control; selective laser sintering

Programme de recherche (Anglais)

COST-Action P4 - Non-linear dynamics in mechanical processing

Description succincte (Anglais)

See abstract

Autres indications (Anglais)

Full name of research-institution/enterprise: EPF Lausanne Laboratoire de gestion et procédés de production STI-IPR

Partenaires et organisations internationales (Anglais)

CZ, DK, F, D, GR, H, IRL, I, NL, PL, SI, E, S, CH, GB

Résumé des résultats (Abstract) (Anglais)

The aim of this project was the control of a manufacturing process named Selective Laser Sintering (SLS) by analysing, thanks to feedforward networks, a process map produced by Finite Element Simulations. We developed a Finite Element Code to compute the temperature evolution inside a part manufactured by SLS. We adapted a criterion to deduce the shape of the part. This criterion is essentially based on a particular processing of the thermal history. We applied our Finite Element Code to two different sintering experiments. In the first experiment, the part to build was a thin parallelepiped made of only one powder layer. In the second experiment, the part was a ring made out of three layers. For each experiments, we identified the operating parameters which could be optimised to improve the process and we decide to represent the part quality by some quantifiable properties which were essentially related to geometry and to mechanical strength. We then run the Finite Element Code with some pre-definite settings of the operating parameters. The output of those simulations were the part properties. We collected the results in a data base and we analysed it with a neural network. In particular, we were able to use this network for finding correct values of the operating parameters leading to a part with desired quality.

Références bases de données (Anglais)

Swiss Database: COST-DB of the State Secretariat for Education and Research Hallwylstrasse 4 CH-3003 Berne, Switzerland Tel. [phone] Swiss Project-Number: C00.0071