基于机器学习的合金强韧化设计研究进展

The synergistic enhancement of strength and toughness is a long-term core challenge in the development of high-performance metallic materials,which involves extremely complex nonlinear coupling relationships among composition,process,microstructure and properties. Traditional trial-and-error approach is inefficient when dealing with a large parameter space. However,the data-driven paradigm represented by machine learning (ML)is accelerating the research and development process of strengthening and toughening alloys. The research progress of ML in solving the problem of alloy strength and toughness inversion is systematically reviewed in this paper. The macroscopic applications of ML in composition,process,and microstructure are reviewed. The combination of ML with strengthening and toughening theories is elaborated. The four basic mechanisms,namely solid solution strengthening,precipitation strengthening,fine grain strengthening,and dislocation strengthening,are analyzed. Furthermore,the applications of microscopic structure image analysis,grain boundary feature extraction,multi-phase collaborative design,and multi-objective design are discussed. Finally,in response to the current limitations in data acquisition and physical mechanisms in research,a future outlook is proposed to build a multimodal data ecosystem and develop interpretable models embedded with physical mechanisms,aiming to provide theoretical references and technical paths for the development of next-generation high-performance strengthening and toughening alloys under extreme service conditions.