EXTENDED ABSTRACT: 7xxx series aluminum alloy is widely used in aviation, aerospace and rail transit fields due to their low density, high strength and good processing performance. It is difficult to design alloy composition and heat treatment process rapidly that meet multi-objective performance requirements only by using trial and error method. The datadriven method developed in recent years provides a feasible way for alloy composition design and heat treatment process optimization. In this work, an efficient alloy design strategy integrating CALPHAD (CALculation of PHAse Diagrams) method, machine learning and key experiment is proposed [1]. Firstly, data reported in the literature are collected, then the effects of alloy composition and heat treatment on mechanical properties are analyzed, subsequently a set of "composition-process-properties" data set of 7xxx series aluminum alloy is established. Based on collected and sorted literature data, the regression analysis of functional relationships between composition and mechanical properties of the alloys was carried out by using BP, RBF and GA-BP neural network models, respectively. The results show that the GA-BP neural network has a higher prediction accuracy than BP and RBF networks. The correlation coefficient R, average absolute relative error AARE, and root mean square error are 0.948,4.28 %, and 0.087, respectively. The multiobjective optimization of mechanical properties was carried out. Meanwhile, SEM, EBSD, HRTEM and tensile test were used to verify experimental results. This work provides a new strategy for the intelligent design of composition and process of high-performance aluminum alloys.
Keywords:aluminum alloy; machine learning; CALPHAD; alloy design
REFERENCES:
[1] B. Li, Y. Du,Z.S. Zheng et al., J. Mater. Res. Technol., 19, (2022) 2483
Yong Du is the Chinese director of the Sino-German cooperation group "Microstructure", and director of Science Center for Phase Diagram & Materials Design and Manufacture, CSU. He was selected to be National Outstanding Youth of National Natural Science Foundation of
China in 2004, Cheung Kong Chair Professorship of Ministry of Education of China in 2006. So far, he has been awarded one First Class Prize of Hunan Provincial Natural Science, one Third Class Prize of National Natural Science. He has published 902 papers with 24145 SCI citation.