Multiobjective Optimization of Heat‐Treatment Pathways for SK85 Steel: Integrating NGBoost and NSGA‐II for Tailored Properties

Information from the abstract

Optimizing the mechanical properties of high‐carbon steels requires precise control over complex phase transformations driven by heat‐treatment pathways. This study develops a novel data‐driven framework integrating natural gradient boosting (NGBoost) for probabilistic surrogate modeling and the non‐dominated sorting genetic algorithm II (NSGA‐II) for multiobjective optimization to design quenching–tempering (Q‐T) and austempering–tempering (A‐T) sequences for JIS SK85 high‐carbon steel. A systematic experimental dataset linking process parameters to microstructural and mechanical responses was constructed. The NGBoost model accurately predicted tensile properties and hardness while quantifying predictive uncertainty, outperforming conventional regression methods. Global sensitivity analysis via Sobol indices identified austempering and tempering temperatures as dominant factors governing mechanical performance. Multiobjective optimization using NSGA‐II revealed that the A‐T route at 320 °C delivers the optimal strength‐ductility balance, attributed to the formation of lower bainite with finely dispersed carbides. K‐means clustering and TOPSIS were employed to select practical Pareto‐optimal solutions. Experimental validation confirmed the model's high accuracy, with errors below 5%. This work demonstrates an effective AI‐aided methodology that reduces trial‐and‐error in designing high‐performance heat treatments for advanced high‐carbon steel components.

Why this record is monitored

This record has an Impact Signal of 72/100 based on recency, source, collaboration, and bibliographic signals. It prioritizes monitoring and is not a judgment of research quality.

Related topics: Microstructure and Mechanical Properties of Steels · Machine Learning in Materials Science · Advanced Multi-Objective Optimization Algorithms