| About this Abstract |
| Meeting |
MS&T21: Materials Science & Technology
|
| Symposium
|
Additive Manufacturing of Ceramic-based Materials: Process Development, Materials, Process Optimization and Applications
|
| Presentation Title |
Machine-learning-based Microstructure-property Prediction Enabled by High-throughput Ceramic Sample-array Preparation Using Integrated Additive/Subtractive Manufacturing |
| Author(s) |
Xiao Geng, Jianan Tang, Dongsheng Li, Yunfeng Shi, Rajendra Kumar Bordia, Jianhua Tong, Xiao Hai, Fei Peng |
| On-Site Speaker (Planned) |
Fei Peng |
| Abstract Scope |
We demonstrate a high throughput sample fabrication method, and a novel machine learning algorithm that can predict the microstructure of laser-sintered alumina. The alumina sample array was prepared using laser-based integrated additive/subtractive manufacturing (IASM) method. We can simultaneously fabricate a sample array that contains hundreds of individual sample units. These sample units have their own microstructure, depending on the laser power distribution and the sample locations. Micro-indentation was carried out to measure the hardness of all the sample units. The microstructure of selected sample units was characterized using SEM. Thus, we efficiently established a database of microstructure and hardness of laser-sintered alumina. We developed a novel machine learning algorithm, regression-based conditional generative adversarial networks (GANs) with Wasserstein loss function and gradient penalty (RCWGAN-GP). We found that RCWGAN-GP can not only accurately regenerate the microstructure of alumina with measured hardness, but also accurately predict the microstructure of alumina of arbitrary hardness. |