EXTENDED ABSTRACT: Nickel-based superalloys are crucial for aviation and gas turbine components such as turbine blades due to their exceptional high-temperature properties, creep resistance, heat resistance, and corrosion resistance. However, optimizing their composition presents challenges due to the extensive compositional space, complex elemental interactions, and potential property conflicts. In this report, a vast dataset encompassing composition, microscopic parameters, and mechanical properties was constructed based on thermodynamic calculations. High-precision prediction models for the relationships between composition, microscopic parameters, and performance were developed using machine learning. By integrating data-driven approaches with domain knowledge and employing constrained multiobjective optimization algorithms, a series of novel nickelbased superalloys with excellent comprehensive performance, including single crystals, directionally solidified crystals, and equiaxed crystals, were efficiently designed. This strategy also enables the creation of superalloys for additive manufacturing, balancing performance and crack resistance, and new refractory high-entropy alloys are designed to achieve high strength and ductility. This design strategy offers a new approach to balancing confficting performance objectives in materials design.

Keywords：Nickel-based superalloys; Data-Driven; Machine Learning; Multi-Objective Optimization