Study on simulation of phase transformation process and prediction algorithm of material properties in metal heat treatment

The phase transformation process of metal heat treatment is complex and plays a decisive role in material properties, so it is of great significance to accurately simulate the phase transformation process and predict material properties. In this paper, a multi-scale simulation framework is proposed, which combines finite element method (FEM) and phase field simulation (PFM) to simulate the evolution of temperature field, stress field and microstructure during metal heat treatment through the coupling of macro and micro scales. At the same time, based on machine learning algorithm, a hybrid neural network model is constructed, and the yield strength, fracture toughness and Vickers hardness of materials are predicted with key heat treatment parameters and microstructure characteristics as inputs. In the experimental part, AISI 4140 alloy steel is taken as the research object, and the accuracy of the multi-scale simulation framework and the predictive ability of the hybrid neural network model are verified through quenching and tempering processes. The results show that the prediction error of microstructure evolution by multi-scale model is less than 6%, and the hybrid neural network model has high accuracy in predicting mechanical properties, and the errors are all less than 6%, but the prediction accuracy of fracture toughness needs to be further improved. This study provides theoretical support and technical means for optimizing heat treatment process and developing new materials.