The determinants of tensile strength in precipitation-strengthened martensitic steels via machine learning

Martensitic steels are widely recognized for their exceptionally high strength, primarily determined by their alloy composition and thermomechanical processing. This study aims to elucidate the influence of these factors on the tensile strength of martensitic steels. To achieve this, datasets have been collected and categorized to accurately reflect the effects of alloy composition. High-accuracy machine-learning models are trained to capture these categorizations, and the contributions of composition and processing parameters to tensile properties are quantitatively assessed. The analysis shows that variations in strengthening mechanisms are tied to distinct precipitated phases, motivating separate models for single-phase martensitic alloy steel (SMAS) and single-phase martensitic carbon steel (SMCS). Subsequently, a genetic programming-based symbolic regression (GP-SR) approach was employed to derive an interpretable formula that is highly correlated with the target performance for both steel types. This work provides valuable insights into quantifying the effects of composition and processing on the mechanical performance of martensitic steels, thereby enabling the optimization and design of martensitic steels with desired strength.