Intelligent optimized design of novel high‐temperature titanium alloys

Abstract 600°C is regarded as the “thermal barrier” temperature for traditional Ti‐based alloys. As the working temperature rises, alloys' creep performance and strength at high temperatures exhibit a dramatic decrease, which becomes a major obstacle to the development of high‐temperature titanium alloys. In order to break the thermal barrier temperature, a new design strategy that integrates machine learning with multiobjective optimization has been employed. A high‐precision predictive model has been established, achieving R 2 values exceeding 0.9, with mean absolute error (MAE) and root mean square error (RMSE) not exceeding 5 and 11, respectively. By referencing domain knowledge, constraints have been proposed, leading to the optimization. Additionally, the α 2 phase is utilized as a reinforcement phase, balancing plasticity while controlling its content range. Titanium alloys that demonstrate high yield strength (greater than 490 MPa) and extended creep life (exceeding 25 h), suitable for conditions up to 650°C, have been designed using multiobjective optimization with constraints. Compared to current typical high‐temperature titanium alloys, these newly developed alloys exhibit superior yield strength and creep life with similar density and cost. This method provides a valuable reference for designing advanced high‐temperature titanium alloys.