Thermodynamic Database and Machine Learning for Heat-Treatment-Free Al-Si-Mg-Cu Alloy Design

This work establishes a thermodynamic database for the Al-Si-Mg-Cu quaternary system, focusing on the Al-Si terminal. Using machine learning methods, preliminary predictions of the complete composition range were conducted, establishing a quantitative relationship between composition, process, and properties. Future research aims to expand the predicted composition range to include Cu content from 0 wt% to 3 wt%, avoiding peak mechanical properties at the composition boundary. The predicted step size will be reduced to 0.1 for greater accuracy. By incorporating 'composition-process-microstructure' as input variables, the entire composition range will be predicted, establishing a quantitative 'composition-process-microstructure-properties' relationship. Optimal alloy compositions will be predicted and screened, followed by key experiments on the Al-Si-Mg-Cu alloy. Mechanical property detection and microstructure analysis will verify the theoretical design's reliability, analyzing the influence of Cu content. Ultimately, this research aims to design a heat-treatment-free Al-Si-Mg-Cu alloy.