About this Abstract |
Meeting |
2025 TMS Annual Meeting & Exhibition
|
Symposium
|
Algorithms Development in Materials Science and Engineering
|
Presentation Title |
Material Characterization for Sheet Metal Forming Processes Using Deep Learning Methods for Time Series Processing |
Author(s) |
Papdo Tchasse, Kim Rouven Riedmüller, Mathias Liewald |
On-Site Speaker (Planned) |
Papdo Tchasse |
Abstract Scope |
The material characterization is the prior step for every product development in the metal forming industry. At this stage, the objective is to determine the different properties of the material, which are significant for the intended manufacturing processes and the final use of the product. Characterizing a material and determining its significant properties can however be a very time consuming and cost-intensive task as it implies that many different experiments are carried out. This paper addresses this issue and proposes a resource and time-efficient approach for material characterization, which is based on deep learning methods for time series processing. For this study, different class of sheet materials were considered and their flow curves, yield locus and forming limit curves were processed. The Methods described in this paper were applied on a data set of about 75 materials and a high correlation score could be attained on the designed regression tasks. |
Proceedings Inclusion? |
Planned: |
Keywords |
Characterization, Machine Learning, Iron and Steel |