PSI - Issue 75

Jörg Baumgartner et al. / Procedia Structural Integrity 75 (2025) 538–545 / Structural Integrity Procedia 00 (2025) 000–000 Jo¨rg Baumgartner

543

6

20 . 0

17 . 5

2 . 5 target max. principal stress [N / mm²] 5 . 0 7 . 5 10 . 0 12 . 5 15 . 0

0 . 0

0

5 20 predicted max. principal stress [N / mm²] 10 15

Fig. 4. Comparison of stress prediction ML to 3D FE model

4 . 0

3 . 5

3 . 0

2 . 5

2 . 0

target r-coord [mm]

1 . 5

1 . 0

1

2

3

4

predicted r-coord [mm]

Fig. 5. Comparison of radial position prediction ML to 3D FE model

6. Conclusions and Outlook

The following main conclusions can be drawn from the investigations:

• The machine learning algorithms are able to reliably estimate notch stresses from cutting forces. This opens up the possibility to use common and fast modeling techniques while utilizing local stresses for further assessment.