Data‐Driven Approach for Simultaneous Optimization of Strength and Conductivity in Al Alloys

Achieving an optimal balance between the mechanical and electrical properties of aluminium alloys through conventional trial‐and‐error methods has posed a significant challenge. This study proposes a machine learning‐driven alloy design strategy by identifying key features through correlation screening, recursive elimination, and exhaustive feature selection techniques, followed by Bayesian optimization (BO) for iterative composition design. Out of various regression models trained, the random forest exhibited the best predictive performance. Based on a comprehensive dataset of aluminium alloys, the most influential factors affecting ultimate tensile strength (UTS) were identified as volume size factor, second ionization energy, and Pauling electronegativity. In comparison, electrical conductivity (EC) was primarily influenced by volume size factor, absolute electronegativity, and work function. Importantly, for the first time, the volume size factor and the electron work function were both identified as key features influencing UTS and EC. Predictive models for UTS and EC were developed, with errors of less than 8% and 9%, respectively. A new aluminium alloy composition, Al‐3.16Cu‐3 Mg‐0.23Cr‐0.21Si‐0.18 Mn (wt.%), was designed through BO and experimentally validated. The alloy showed an excellent combination of properties with a UTS of 278 MPa and EC of 52.68% international annealed copper standard (IACS), demonstrating the effectiveness of the proposed method in simultaneously improving mechanical and electrical performance.