Akademik Veri Yönetim Sistemi
Engineering Science and Technology, an International Journal, cilt.77, 2026 (SCI-Expanded, Scopus)
Controlling grain growth during high temperature processing remains a critical challenge in the development of titanium alloys, as excessive β-grain coarsening severely degrades mechanical properties. Although machine learning has increasingly been applied in materials science, its use for the systematic and mechanism-informed suppression of β-grain growth in titanium alloys has not been previously reported. In this study, a machine learning–guided framework is presented for the first time to design titanium alloys with intrinsic resistance to grain growth by integrating data-driven modelling, robust optimization, and kinetic analysis. A comprehensive database was constructed by correlating alloy composition, processing temperature, and holding time with β-grain size evolution. Pearson correlation analysis was employed to identify the most influential parameters governing grain growth behaviour. Multiple regression algorithms were evaluated to establish accurate surrogate models, among which the Gradient Boosting Regressor demonstrated superior predictive accuracy and robustness. Building on the trained surrogate models, a robust Bayesian optimization strategy was implemented to identify alloy compositions that minimize grain growth across a wide temperature–time window. The optimized Ti–Nb–Mo–Zr–Ta–Sn alloy exhibited consistently suppressed β-grain coarsening compared to conventional titanium alloys. Grain growth kinetics analysis revealed diffusion-controlled behaviour characterized by a high grain growth exponent and a large apparent activation energy of 961.10 kJ·mol−1, indicating sluggish grain boundary migration dominated by solute drag effects. Furthermore, thermodynamic calculations confirmed the stability of a single β phase within the investigated processing range, supporting the metallurgical feasibility of the optimized alloy. Overall, this work demonstrates that integrating machine learning, optimization, and kinetic analysis provides a powerful and generalizable strategy for microstructure tailoring and the design of thermally stable titanium alloys.