Machine learning assisted prediction for the coefficient of thermal expansion of binary crystals
Hongyu Yang, C. Gao, Denghui Jiang, Dafang Zhong, Yuxuan Ma, Yang Li, Linzhuang Xing, Heng Zhao, Yang Li, Zhimin Li, Yue Hao
Abstract
It is challenging to theoretically predict the coefficient of thermal expansion (CTE) for binary crystals A<sub>m</sub>B<sub>n</sub> owing to the complexities of crystal structures and computational procedures. Herein, the Pearson feature selection method is utilized to identify nine key features associated closely with crystal structures, and a back-propagation neural network model with two hidden layers containing 24 and 15 neurons is adopted to achieve the optimal matching effect of CTE, which is specially optimized by the pelican optimization algorithm. Moreover, the black-box nature of the model is well elucidated by interpretability techniques of SHapley Additive exPlanations and Accumulated Local Effects, including the specific impact rules of each feature and the interaction effects between features on CTE. It is found that the feature of average bond length delivers the contribution of up to 27%, while low influence features serve an important function in increasing prediction accuracy. The findings demonstrate high efficiency and accuracy of the developed model for predicting CTE of binary crystals.