| About this Abstract |
| Meeting |
MS&T21: Materials Science & Technology
|
| Symposium
|
Advanced Characterization of Materials for Nuclear, Radiation, and Extreme Environments
|
| Presentation Title |
Utilizing a Dynamic Segmentation Convolutional Neural Network for Microstructure Analysis |
| Author(s) |
Stephen Taller, Luke Scime, Kurt Terrani |
| On-Site Speaker (Planned) |
Stephen Taller |
| Abstract Scope |
Microstructural features such as precipitates, grain boundaries, and dislocations will dictate the mechanical properties of materials in extreme environments. This presents a significant challenge to use analytical electron microscopy to characterize the size, number density, composition, and volume fraction of each microstructural feature. The ultimate objective of this work is to demonstrate a generalized approach for characterization of complex microstructures at large scale. This approach is now possible through (1) advances in microscopy and its automation to assess large areas of material, and (2) advances in pixel-wise machine learning classification tools that are agnostic to the image size, type, and number of input channels. This presentation focuses on the application of a dynamic segmentation convolutional neural network for rapid microstructural analysis as demonstrated on an additively manufactured sample of Ni-superalloy 718. This generalized approach will provide for detailed microstructural characterization to facilitate more accurate property predictions. |