A data-driven approach for automating the design process of deep drawing tools
Publication date
2025
Document type
Konferenzbeitrag
Author
Organisational unit
Conference
13th International Conference and Workshop on Numerical Simulation of 3D Sheet Metal Forming Processes ; Munich, Germany ; July 7–11, 2025
Publisher
IOP Publishing
Series or journal
Journal of Physics: Conference Series
ISSN
Periodical volume
3104
Periodical issue
1
Peer-reviewed
✅
Part of the university bibliography
✅
Language
English
Abstract
The deep drawing tool development process, from method planning and design of tools to tool try-out and final commissioning, is very time-consuming and requires extensive iterative manual effort, particularly during the try-out stage. To accelerate the entire process, integrating obtained knowledge from the tool try-out stage into the early design stage offers significant potential. Towards automating tool design, this paper proposes a data-driven approach using a generative neural network to predict active surfaces of deep drawing tools based on given deep drawn parts, laying the foundation for incorporating try-out knowledge. The model is trained on active tool surfaces and their corresponding deep drawn parts, including variation of geometrical parameters and process parameters in deep drawing simulation. The approach is evaluated using simulated data from deep drawing processes. The proposed solution demonstrates an advancement in automatically generating the active tool surfaces for both the punch and the die directly from the desired deep drawn parts.
Version
Published version
Access right on openHSU
Metadata only access