Machine Learning Prediction on Properties of Nanoporous Materials Utilizing Pore Geometry Barcodes

• DOI: 10.1021/acs.jcim.9b00623
• Publication Date: 2019-10-30T01:06:57Z
• Authors: Xiangyu Zhang; Jing Cui; Kexin Zhang; Jiasheng Wu; Yongjin Lee
• Keywords: Nanoporous

In this work, we propose a computational framework for machine learning prediction on structural and performance properties of nanoporous materials methane storage application. For our prediction, two descriptors based pore geometry barcodes were developed; one descriptor is set distances from structure to the most diverse in barcode space, second extracts uses important features barcodes. First, identify optimal condition effects training preparation method, size, models investigated. Our analysis showed that kernel ridge regression provides highest accuracy, randomly selected 5% structures entire would work well as set. results both accurately predicted even zeolites. Furthermore, demonstrated approach predicts metal–organic frameworks, which might indicate possibility be easily applied predict other types materials.