About this Abstract |
Meeting |
MS&T23: Materials Science & Technology
|
Symposium
|
Advances in Ferrous Metallurgy
|
Presentation Title |
Accurate Classification of Bainitic and Tempered Martensitic Steels with Advanced Deep Learning Methods |
Author(s) |
Xiaohan Bie, Juancheng Li, Manoj Arthanari, Evelin Barbosa de Melo, Jun Song, Steve Yue |
On-Site Speaker (Planned) |
Xiaohan Bie |
Abstract Scope |
Microstructures are vital for mechanical behaviors of high-strength steel (HSS). Lower bainite (LB) and tempered martensite (TM) are two common microstructures in HSS. Though previous studies attempted to describe the differences between LB and TM, in situations when there is no prior knowledge of the heat treatment, it remains very challenging to differentiate the two directly from the characterization (e.g., microscopy) images, particularly for untrained eyes. Deep Learning methods have been shown to surpass human performance in classification tasks. Here we employed deep Learning to classify microstructure images of LB and TM, and achieved an impressive 96.81% accuracy in differentiating LB and TM. Our investigation of the learning process revealed significant similarities in their microstructures, indicating a well-defined processing difference but not a microstructural difference between the two. Our findings demonstrate the potential of deep Learning in distinguishing challenging microstructures and advancing our comprehension of HSS microstructures. |