About this Abstract |
Meeting |
2020 TMS Annual Meeting & Exhibition
|
Symposium
|
Additive Manufacturing: Advanced Characterization with Synchrotron, Neutron, and In Situ Laboratory-scale Techniques
|
Presentation Title |
In-situ Process Monitoring and Diagnosis via Machine Learning of Thermal Imaging Streams |
Author(s) |
Linkan Bian |
On-Site Speaker (Planned) |
Linkan Bian |
Abstract Scope |
One current challenge in Laser Based Additive Manufacturing (LBAM) is the potential defects and structural integrity of fabricated parts. To improve quality of fabricated parts, accurate predictions of part quality are needed. We develop a novel machine learning framework that accurately predicts the physics and mechanical properties well within LBAM tolerance limits by considering the local heat transfer. The central hypothesis is that terabytes of thermal imaging data generated during the LBAM fabrication is highly correlated to the relevant part properties, and thus can be used to predict the distribution of internal defects and external geometric characteristics. To achieve this goal, efficient methodology is needed to extract relevant features based on big thermal data. Our tensor-based machine learning approach not only gives highly accurate NDE for the fabricated parts but also provide a realistic approach for handling big data generated from parts with large size and complex geometries. |
Proceedings Inclusion? |
Planned: Supplemental Proceedings volume |
Keywords |
Advanced Processing, |