| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
High Performance Steels
|
| Presentation Title |
Data-Driven Prediction Model for Surface Hardness Distribution in Nitrided Steel |
| Author(s) |
Goro Miyamoto, Sayaka Sekida, Tadashi Furuhara |
| On-Site Speaker (Planned) |
Goro Miyamoto |
| Abstract Scope |
To achieve carbon neutrality, extending the lifespan of gears in electric vehicles and wind turbines is essential. Surface hardening treatments are effective in improving fatigue strength. Nitriding, one of the popular surface hardening treatments, involves nitrogen absorption and diffusion. The hardness distribution after nitriding is affected by processing conditions and alloying elements. Predicting hardness distribution is crucial for optimal processing and alloy design of nitrided steels. Therefore, our group developed a data-driven model to predict hardness distribution. We built a dataset of more than 800 conditions using experimental and data-mining from literature. Using this dataset, a deep learning model was created to predict hardness distribution with explanatory parameters such as alloy composition, nitriding temperature, time, and metallurgical parameters. Comparisons between model predictions and experimental values show accurate reproduction of hardness distribution, demonstrating the effectiveness of our data-driven approach in predicting the effects of various elements and conditions on nitrided steels. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Iron and Steel, Surface Modification and Coatings, Machine Learning |