About this Abstract |
Meeting |
MS&T23: Materials Science & Technology
|
Symposium
|
Materials Processing and Fundamental Understanding Based on Machine Learning and Data Informatics
|
Presentation Title |
Prediction of the Mechanical Response of Zirconia-reinforced Metal-matrix Composite Using Deep Learning Approaches |
Author(s) |
Maryam Shakiba, Marwa Yacouti |
On-Site Speaker (Planned) |
Maryam Shakiba |
Abstract Scope |
The prediction of the stress distribution in structures and materials subjected to external loads is a crucial step for structural design and optimization. Finite element analysis is computationally expensive when applied to complex geometries and non-linear problems. Thus, we propose an accurate and more efficient surrogate for FEA using deep learning methods to capture the evolution of the phase transformation in zirconia particles. A convolutional neural network framework is used to predict the distribution of the mechanical responses of zirconia-reinforced metal-matrix composite subjected to compression. Zirconia can be characterized by two unique properties: superelasticity and the shape memory effect that are achieved thanks to the reversible phase transformation from austenite to martensite. We predict the distribution of the stress and the martensite volume fraction in the zirconia-reinforced metal-matrix composite. Furthermore, we explore the capacity of the deep learning framework to generalize for different zirconia volume fractions and particles’ diameters. |