About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Forming and Joining of Advanced Sheet Metal Materials
|
| Presentation Title |
AI-Based Inverse Process Design for Shear Cutting of Advanced High-Strength Steels |
| Author(s) |
Marcel Goerz, Kim Rouven Riedmüller, Mathias Liewald, Adrian Schenek |
| On-Site Speaker (Planned) |
Marcel Goerz |
| Abstract Scope |
Shear cutting is one of the most used processes in sheet metal forming. A critical quality attribute is the geometry of the cutting surface, which is predominantly determined during the design process. Traditionally, this process has relied on iterative procedures or the tacit knowledge of staff. These conventional methods follow a forward design approach, whereby tool parameters are defined first and the resulting quality of the cut surface is subsequently evaluated. This trial-and-error methodology is time-consuming and ineffective for achieving consistent surface quality, particularly with advanced high-strength steels that exhibit complex material behaviour. This study introduces an inverse design methodology using AI-based surrogate modelling to overcome these limitations. Rather than starting from tool parameters, the proposed approach predicts suitable punch design parameters based on desired cutting surface characteristics. Various AI-based surrogate models are trained and evaluated using a dataset containing surface quality metrics and corresponding punch cutting-edge design parameters. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Iron and Steel, Other |