Figure 6.

Flow chart of high-efficiency alloy design by integrating computational thermodynamics, machine learning, and key experiments. The thermodynamic database was validated by key experiments and can provide feature variables for machine learning through the computation of the high-throughput platform Malac-Distmas, while the mechanical properties obtained from key experimental tests provide training data for machine learning. After training by machine learning, the points that may have the best comprehensive mechanical properties are recommended by the acquisition function (EI) and validated by key experiments. Finally, combining computational thermodynamics, machine learning, and key experiments establishes the relationship between the “composition-process-microstructure-properties” of alloys to achieve efficient alloy design.