What can machine learning help with microstructure-informed materials modeling and design?

TL;DR

This paper reviews how machine learning enhances microstructure-informed materials modeling, covering characterization, simulation, and design, and discusses future research directions to advance this interdisciplinary field.

Contribution

It provides a comprehensive overview of current machine learning applications in microstructure modeling and offers guidance for future research in this rapidly evolving area.

Findings

Summarizes recent machine learning techniques in microstructure characterization.

Highlights successful applications in multiscale simulation and property prediction.

Suggests future research directions and educational resources for interdisciplinary collaboration.

Abstract

Machine learning techniques have been widely employed as effective tools in addressing various engineering challenges in recent years, particularly for the challenging task of microstructure-informed materials modeling. This work provides a comprehensive review of the current machine learning-assisted and data-driven advancements in this field, including microstructure characterization and reconstruction, multiscale simulation, correlations among process, microstructure, and properties, as well as microstructure optimization and inverse design. It outlines the achievements of existing research through best practices and suggests potential avenues for future investigations. Moreover, it prepares the readers with educative instructions of basic knowledge and an overview on machine learning, microstructure descriptors and machine learning-assisted material modeling, lowering the interdisciplinary hurdles. It should help to stimulate and attract more research attention to the rapidly growing field of machine learning-based modeling and design of microstructured materials.