Micro-alloyed steels are extensively used in many industries owing to their high strength, formability, wear and fatigue resistance. Understanding hot flow behavior of these steels under complex thermomechanical conditions is of great significance for manufacturing process optimization. The current work investigates the hot torsion behavior of Ti-Nb micro-alloyed steel at temperatures from 850 to 1100 ∘C and strain rates from 0.01 to 1 . However, existing machine learning (ML) models often lack the robustness and generalizability needed to predict flow stress across diverse processing parameters accurately. To address this limitation, this study introduces a novel stacking machine learning (SML) model for high-precision prediction of flow behavior in Ti-Nb microalloyed steel during hot torsion deformation. The SML model integrates multiple diverse base learners and a meta-modeling strategy, enhancing prediction accuracy, capturing complex non-linear interactions, and mitigating overfitting. The model's performance was evaluated on a comprehensive dataset of 47,960 instances (derived from 12 hot torsion tests) and validated using a 5-fold cross-validation technique. Statistical results demonstrate the superior accuracy of the proposed SML model compared to other models, achieving = 0.999964, MAE= 0.103796, MSE= 0.055507, and RMSE= 0.235599, on the testing dataset. Feature importance analysis using Shapley additive explanations revealed that temperature and strain are the most influential features affecting the target stress value. These findings highlight the effectiveness of the proposed SML model, in accurately predicting the hot flow behavior of micro-alloyed steels. The primary contribution of this work is the development of a highly accurate and generalizable SML model that can be used to optimize hot forming processes for Ti-Nb microalloyed steel, reducing the need for costly and time-consuming trial-and-error experiments.
