About this Abstract |
Meeting |
2025 TMS Annual Meeting & Exhibition
|
Symposium
|
Steels in Extreme Environments
|
Presentation Title |
Segmentation of Microscopy Images of Lower Bainite and Tempered Martensite High Strength Steels |
Author(s) |
Jun Song, Xiaohan Bie, Manoj Arthanari |
On-Site Speaker (Planned) |
Jun Song |
Abstract Scope |
The susceptibility of high-strength steel (HSS) to hydrogen embrittlement (HE) has been demonstrated to be strongly dependent on the microstructure therein. It has been reported that lower bainite (LB) HSS shows more resistance to HE than tempered martensite (TM) HSS, attributed to difference in the distribution and morphology of carbide precipitates. However, comparison of carbides in LB and TM have been mainly qualitative, and heavily relying on localized features. Employing deep learning, this study established a workflow for segmentation of microstructure images, quantitatively analyzing different metrics of carbides in LB and TM HSS of similar strength. Our results revealed that carbides in LB and TM, overall, show great similarity, but exhibit marginal difference in certain characteristics. The deep learning model achieved an accuracy of 97.38%, providing a time-efficient workflow for analyzing carbides in HSS and also promising a potential route toward efficient and quantitative microstructural analysis in structural metals. |
Proceedings Inclusion? |
Planned: |
Keywords |
Iron and Steel, Machine Learning, Characterization |