About this Abstract |
Meeting |
2022 TMS Annual Meeting & Exhibition
|
Symposium
|
Dynamic Behavior of Materials IX
|
Presentation Title |
Machine Learning Based Approach to Modeling and Predicting Material Behavior and Failure Criteria in Composites |
Author(s) |
Tyler Dillard, Nolan Lewis, Abhijeet Dhiman, Vikas Tomar |
On-Site Speaker (Planned) |
Tyler Dillard |
Abstract Scope |
The complex inhomogeneous nature of composites raises particularly difficult challenges when trying to understand, model, and predict the material behavior of composites under load. This work aims to prove the viability and potential benefits of machine learning and other data science-based approaches when predicting material behavior and failure criteria in composites. Composite samples with different microstructures were simulated to indentify different failure scenarios under tensile loading. These failure scenarios serve as a high-fidelity data to train machine learning model. After multiple regression techniques were utilized to trim the data, the data was then fed to a multi-fidelity neural network with machine learning in order to produce crack propagation and failure scenario predictions. These predictions are then compared to the corresponding low fidelity, high fidelity, and mixed fidelity FEM simulation data. |
Proceedings Inclusion? |
Planned: |
Keywords |
Machine Learning, Composites, Mechanical Properties |