Multi-model integration accelerates Al-Zn-Mg-Cu alloy screening

The 7xxx (Al-Zn-Mg-Cu) alloys were extensively utilised in aerospace and rail transit due to their low density, high strength, and excellent processability. Nevertheless, an increase in strength inevitably comes at the expense of ductility, and vice versa, which was known as the “strength-ductility trade-off” dilemma. Extensive research had been conducted to address this issue, accumulating a large amount of experimental and computational data. However, the conventional approach of trial-and-error largely relied on the empirical knowledge of researchers. If the expected results are not obtained, the experiment will need to be repeated constantly. The emergence of machine learning (ML) as a new paradigm had introduced novel data analysis tools for basic scientific research. Numerous studies had demonstrated the feasibility and reliability of ML methods. Consequently, this paper adopted ML methods to investigate this dilemma of Al-Zn-Mg-Cu alloys. A multi-algorithm integrated model was employed to accelerate the screening of the target mechanical properties. Four high-strength and high-ductility alloys were designed using the desired target properties as inputs. The alloys designed by the ML methods were experimentally prepared. Additionally, the effects of extrusion and rolling processes on the alloy’s properties were compared. Notably, the E4 alloy exhibited an ultimate tensile strength (UTS) of 709 ± 4 MPa and an elongation (EL) of 16% ± 1%, which represents a significant enhancement in comprehensive performance. This study provides a reference for resolving the “strength-ductility trade-off” dilemma, and contributes to the development of more competitive aluminium alloy materials.