Application of Machine Learning in the Study of Creep Properties of Alloys

ABSTRACT

Creep resistance is a critical performance indicator for alloy components operating under extreme conditions such as high temperature, high pressure, and long‐term loading, directly affecting operational reliability in key sectors, including the aerospace, nuclear power, and deep‐sea sectors. Traditional research on creep resistance relies on long‐term experimental testing and empirical models, which are limited by long cycles, high costs, and weak generalization capabilities. In recent years, machine learning (ML) has achieved breakthrough progress in areas such as alloy creep performance prediction, the design of new creep‐resistant materials, and component life assessment by leveraging its powerful capabilities in nonlinear mapping, big data processing, and small‐sample learning. This review summarizes the existing literature and systematically explores the application efficacy of ML in predicting the creep life, creep rate, creep deformation, rupture stress, residual life, and creep–fatigue interaction for various alloy systems, including nickel‐based superalloys and titanium alloys. This review focuses on analyzing key technical aspects, such as feature engineering and model optimization, and summarizes the practical value of ML in predicting the life of engineering components and accelerating the design of new creep‐resistant materials. ML provides a novel, efficient, and precise technical pathway for the study of alloy creep properties, driving the relevant fields toward an intelligent transformation from “experiment‐driven” to “data‐physics dual‐driven.”