Establishing the complex, non-linear relationship among composition, processing, microstructure, and performance poses significant challenges for traditional methods. In this study, we propose a machine learning (ML) framework for the rapid prediction of yield strength and inverse composition design of γ′-strengthened nickel-based superalloys. A high-precision non-linear ML model was developed, achieving a coefficient of determination (R²) of 0.95, with a prediction error in the test set of less than 5%. Utilizing a simulated annealing algorithm (SA) for inverse design, we successfully developed a novel high-performance superalloy within the Ni-Co-Cr-Al-Ti system. Experimental results indicate that the designed alloy exhibits a yield strength of 920 MPa and an ultimate tensile strength of 1285 MPa at room temperature, surpassing the upper strength limits of conventional commercial counterparts. Furthermore, microstructure characterization and strengthening calculations reveal that the combination of fine grains (4.62 μm) and a high volume fraction (~70%) of γ′ phases contributes significantly to the alloy's exceptional strength. This methodology represents an efficient data-driven approach for accelerating the design of advanced superalloys.
Zhou et al. (Sun,) studied this question.