Machine Learning‐Enabled Inverse Design of Martensitic Stainless Steels With Enhanced Hardness and Wear Resistance

This study establishes a machine learning‐driven predictive model (MLP‐LOA) to decode the quantitative relationships between alloy composition, heat treatment parameters, and key performance metrics—hardness and specific wear rate—in martensitic stainless steels. The model demonstrates exceptional predictive accuracy (R ² ‐ HV = 0.98603, R ² ‐ SWR = 0.97054). Leveraging this model, a multi‐objective evolutionary algorithm was integrated to achieve synergistic optimization of hardness and specific wear resistance, yielding an optimal solution characterized by superior comprehensive properties. For this optimal performance combination, the model identified the corresponding optimal alloy composition and heat treatment process parameters. Based on the model's optimization, the optimal configuration was experimentally established as C = 0.23 wt%, Cr = 13.59 wt%, combined with quenching at 990°C and tempering at 200°C. Experiments were then designed and conducted based on this optimal composition and process scheme. The experimental results show excellent agreement with the predictions of the model (hardness deviation < 3%, specific wear rate deviation < 10%), robustly validating the efficacy of this target‐performance‐driven inverse design framework for performance‐oriented alloy development.