A review on Machine learning aspect in physics and mechanics of glasses

Abstract

The glass science and technology is a rapidly developing field which is focused on development of new glasses with excellent properties. Glasses are the non-crystalline materials with inherent stoichiometry i.e. non-disordered structure of atoms and molecules, thus inherently unpredictable. The ineffective trial-and-error methods are typical to glasses design. The classical computational methods such as ab initio and classical molecular dynamics simulation techniques are costly, time consuming and provide limited data of results. To overcome from such problems, the machine learning (ML) replaces the classical experimental and simulation techniques to produce results more precisely. In the recent years, a lot of studies are carried out on AI to develop new compositions of glasses based on the different types of input parameters. Researchers developed new glasses by improving the various properties of glass like edge strength, shear strength, tensile strength, delamination, etc. In this paper, an effort has been made to explore recent developments in glass manufacturing and technology by the implementation of ML techniques. In this paper, the development of glass of new composition, prediction of glass properties, and various inspection methods are discussed on the basis of application of ML techniques. © 2022 Elsevier B.V.