Masterarbeit | Fachbereich F4; F1; F1.1

Artificial Neural Network based Geometric Compensation for Thermal-induced Deformation in LPBF

Unser Profil

The Chair for Digital Additive Production DAP at RWTH Aachen University investigates the fundamental technical and economic interrelationships of additive production together with partners from industry and science.
In addition to the further development of existing AM processes and existing machine and system technology, the focus on software-driven end-to-end processes is an essential working point of the DAP. Starting with bionic lightweight construction, through functional optimization for AM and the design of digital materials, to validation in the real process and the derivation of static and dynamic characteristic values, the advantages of additive processes can be utilized using digital technologies. Almost all common software suites in the area of authoring systems (CAD) and commercially available CAx systems, FEM modelers, etc. are available for this purpose. On the machine side, both commercially available systems and adapted laboratory systems and test setups are available.

Dein Profil

You are a highly motivated individual and interested in shaping the future of 3d printing with us?!
Prior knowledge in programming and/or metal 3d printing is advantageous (but not required).

Deine Aufgaben

The goal of this thesis is to use an artificial neural network (ANN) based methodology to modify the geometry of the part in CAD software products such that after the processing, the produced part would be close to the expected geometry. The algorithms need to learn from the geometric errors between the CAD model and the L-PBF-produced part and to predict the needed compensation for the similar parts. The modifications is applied in the model and then, the revised model will be converted to an STL file. The other option is to apply the changes directly in the STL file. To compare the geometry of the produced part with its CAD model, 3D scanning techniques will be used. It is worth mentioning that it is not meant to develop a neural network algorithm in this thesis, but the goal is to mostly focus on the existing ones and to use them for the described purpose. Within the scope of this work, the applicability of the algorithm will be illustrated and a demonstrator will be manufactured to validate the capability of the algorithm.

Haben wir Dein Interesse geweckt?

Sende uns bitte Deine aussagekräftigen Bewerbungsunterlagen mit Lebenslauf, Abschlusszeugnissen und aktuellen Studienleistungen unter Angabe der Kennziffer 421510-1764 per Email an folgende Adresse:

Jonas Zielinski, M. Sc.

RWTH Aachen Lehrstuhl für
Digital Additive Production DAP
Steinbachstraße 15
52074 Aachen

→ E-Mail senden

Weitere Ausschreibungen