Data driven structure property relationship and rational design of rare-earth magnetostrictive materials

Rare-earth magnetostrictive materials possess irreplaceable strategic value in fields such as national defense, precision actuation, and smart sensing. Conventional materials development relies heavily on experimental trial-and-error approaches, first-principles calculations, and numerical simulations. However, owing to the complexity of multiphysics coupling, weak cross-scale correlations, and high computational cost, it is difficult to achieve both high efficiency and high accuracy. This paper systematically reviews recent progress in the application of machine learning (ML) to rare-earth magnetostrictive materials. From electronic structure and microstructural evolution to macroscopic property responses, the review clarifies the physical information transmission pathways between different scales. Furthermore, the challenges and development opportunities faced by cross-scale information fusion for machine learning-assisted design of magnetostrictive materials are discussed. ML methods are then classified and summarized according to different research scenarios, including property prediction, composition optimization, structure-property mapping, and parameter design for rare-earth magnetostrictive materials. In addition, this review further analyzes several core issues currently faced in this field and provides an outlook on future developments. This review provides a reference for the integration of data-driven methods and physical mechanisms, as well as for the design of high-performance magnetostrictive materials.