神经网络在金属塑性本构模型方面的研究进展

The accurate construction of constitutive models is crucial for ensuring the safety, reliability and optimal design of structures, and directly affects the material selection, manufacturing processes and product performance. The neural network models, as a new phenomenological constitutive model, are widely used to characterize and predict the nonlinear behavior under complex loading conditions. This paper systematically presented the research progress of neural networks in multigrain microscopic mechanics, hardening models, yield equations, fracture criteria and forming limits, analyzed the problems existing in related studies and briefly described the application prospect of neural networks in metal plasticity. It discussed the aspects requiring further in-depth research in material mechanics, structural design and engineering applications, aiming to promote research progress both domestically and internationally.