Efficient Machine Learning Optimization of Gear Micro Geometry and Comparison with Manually Designed Gears
- Delivery
- Available on this site
- Format
- Price
- Non-members (tax incl.)：￥1,100 Members (tax incl.)：￥880
- Publication code
- 20265027
- Paper/Info type
- Proceedings (Spring)
No.6-26
- Pages
- 1-8(Total 8 p)
- Date of publication
- May 2026
- Publisher
- JSAE
- Language
- English
- Event
- 2026 JSAE Annual Congress (Spring)
Detailed Information
| Author(J) | 1) Andrew Wild, 2) Simon Terry, 3) Paul Langlois, 4) Jaekwon Lee, 5) Shinya Sudo |
|---|---|
| Author(E) | 1) Andrew Wild, 2) Simon Terry, 3) Paul Langlois, 4) Jaekwon Lee, 5) Shinya Sudo |
| Affiliation(J) | 1) SMT, 2) SMT, 3) SMT, 4) Hino Motors, 5) Hino Motors |
| Affiliation(E) | 1) SMT, 2) SMT, 3) SMT, 4) Hino Motors, 5) Hino Motors |
| Abstract(E) | Optimisation of gear micro geometry guided by machine-learnt surrogate models is compared to a past engineer-led optimisation and proven to be a highly efficient technique. Subject to engineer-defined objectives, the optimiser identifies high-performing micro geometries with no further human intervention, with all predictions automatically validated by a hybrid Hertzian and FE-based loaded tooth contact analysis. Going beyond the scope of the original manual optimisation, the optimiser is used to identify micro geometries that are torque robust or retain similar performance to the specification without using different micro geometries on the two input gears, thereby theoretically reducing manufacturing costs. |
