A machine learning multitasking integrated framework and its derived [Mo]eq-[Fe]eq map for designing TRIP titanium alloys

• A high-precision multitasking integrated framework for TRIP titanium alloy design was developed. • The integrated framework is interpretability, enabling the identification of key influencing factors for the TRIP effect. • First proposed, the [Mo] eq -[Fe] eq map enables quick judgement of TRIP effect occurrence in titanium alloys. To enable rapid and accurate design of TRIP titanium alloys, an integrated multitasking framework based on the Stacking algorithm was developed, for the first time integrating the Sparrow Search Algorithm (SSA) to optimize hyperparameters in the framework. The SSA-Stacking framework exhibited satisfactory predictive accuracy on the test set and outperformed traditional base models. A TRIP titanium alloy Ti-5Mo-5 V-3Cr-3Al-2Nb (wt. %) was designed based on the SSA-Stacking framework. It was experimentally verified that the TRIP effect occurred after the alloy was solution-treated at 870 °C for 0.5 h, which in turn achieved a good strength-plasticity matching with a tensile strength of 864 MPa, a yield strength of 748 MPa, and an elongation of 28 %. To provide researchers with a more convenient semi-empirical method for designing TRIP titanium alloys, a [Mo] eq -[Fe] eq map was constructed based on the dataset, SSA-Stacking framework, and SHAP analysis. Researchers can directly utilize the [Mo] eq -[Fe] eq map established in this study to rapidly determine whether the TRIP effect can occur. The [Mo] eq -[Fe] eq map demonstrates higher accuracy than the traditional semi-empirical design methods. Further combined with interaction analysis between [Mo] eq , [Fe] eq , and solution temperature, this study provides quantitative guidance for optimizing the chemical composition and heat treatment processes in TRIP titanium alloy development.