A preliminary discussion about the application of machine learning in the field of constitutive modeling focusing on alloys

With an emphasis on the development of machine learning-based constitutive modeling approaches, the state of constitutive modeling techniques and applications for metals and alloys was examined. This study explored three distinct methods of constitutive modeling: phenomenological constitutive models, physical-based constitutive models, and machine learning-based constitutive models. There was discussion of the benefits and drawbacks of three constitutive models. Analyzed and reviewed were the corresponding uses of neural networks in phenomenological models and physics-based constitutive models. In-depth analysis was conducted on the development of machine learning-based constitutive modeling methods and their uses, particularly the application of machine learning in combination with finite elements (FE). The trends of constitutive modeling approaches are finally covered after going over and summarizing the advantages and current trends of employing machine learning modeling techniques in place of numerical constitutive modeling techniques. The review may offer a more comprehensive reference for the advancement of constitutive modeling techniques for metallic materials research.