About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Materials Informatics to Accelerate Nuclear Materials Investigation
|
Presentation Title |
Accelerating Characterization of Radiation Driven Processes using Machine Learning Tools |
Author(s) |
Stephen Taller |
On-Site Speaker (Planned) |
Stephen Taller |
Abstract Scope |
Over the last half century, tools such as the electron microscope enabled insight into processes surrounding radiation induced defects and microstructure evolution. State-of-the-art electron microscopes can acquire large image datasets rapidly across multiple detectors and spectroscopy modes. Manual annotation analysis of high-resolution data is inefficient and may no longer be feasible in the era of big data. Additionally, expert identification still leads to a large variation in quantification. This presentation will focus on the development and application of a pixel-wise defect detection machine learning tool called a dynamic segmentation convolutional neural network with associated automated image acquisition and post-processing to rapidly quantify information in both static and dynamic (video) transmission electron microscopy datasets. Examples focus on precipitate, cavity, and dislocation evolution in superalloy 718 and include microstructures pre-irradiation, evolving during in-situ ion irradiations, and post-irradiation in the High Flux Isotope Reactor at nominally 300°C or 600°C. |
Proceedings Inclusion? |
Planned: |
Keywords |
Characterization, Machine Learning, Nuclear Materials |