Accelerated design of L12-strengthened Co-base superalloys based on machine learning of experimental data
Jinxin Yu, Chenglei Wang, Chenglei Wang, Yuechao Chen, Cuiping Wang, Cuiping Wang, Xingjun Liu
Abstract
Co-base superalloys strengthened by γ′ precipitates have been regarded as a candidate for aircraft engines. However, its γ′ precipitates are not stable. Moreover, improving its properties experimentally could spend a lot of time. DFT and thermodynamic calculation also have disadvantages, which could not accelerate the designing process of Co-base superalloys significantly. Thus, a new strategy is needed to predict the properties of the superalloys rapidly and accurately. In this study, an accelerated design strategy is applied to find the Co-base superalloys with good properties. Four important properties, which are the existence of γ' and other phases, the γ' solvus temperature and area fraction are predicted based on machine learning-based models. Samples used in this study are experimental data collected from related references and our previous study. Afterwards, four predicting models are integrated to design the superalloys that meet the designing requirements of four properties simultaneously. Finally, six groups are chosen from 363,000 possible candidates and all of six are experimental validated. New Co-base superalloys with high γ' solvus temperature and high γ' area fraction are designed successfully. Our strategy is suitable for the rapid multi-properties design of other advanced materials.