基于机器学习的面心立方高熵合金弹性性能预测

Prediction of elastic properties of face-centered cubic high-entropy alloys by machine learning

提出预测面心立方高熵合金弹性性能的机器学习模型。数据集来源于第一性原理计算得到的186组样本。测试集体积模量(B)和剪切模量(G)预测值的拟合优度(R2)分别达到0.81和0.84。根据机器学习结果,Co Ni CuMoW在等主元高熵合金(G/B≤0.57)中具有最高的B、G和弹性模量(Y)和良好的塑性。第一性原理计算结果表明,当W含量增加时,(CoNiCuMo)1-xWx合金的弹性各向异性提高,且塑性下降。差分电荷密度分析结果表明,W—W和W—Mo键存在明显的电荷聚集,表明W原子与邻近原子间存在共价键作用。

The machine learning(ML)models were proposed for predicting elastic properties of face-centered-cubic(FCC)high-entropy alloys(HEAs).The data set was from the first-principles calculation,which contained 186 samples.The goodness-of-fit(R~2)values of predicted bulk modulus(B)and shear modulus(G)in the test set were 0.81 and 0.84,respectively.According to the results of ML,Co Ni Cu Mo W HEAs have the largest B,G,elastic modulus(Y)and good ductility(G/B≤0.57)among the FCC HEAs with equal components.The first-principles calculation results show that the elastic anisotropy of(Co Ni Cu Mo)1-xWx HEAs increases and ductility decreases with increasing W content.According to the analysis of charge density difference,there is obvious charge accumulation at W—W and W—Mo bonds,indicating the directional covalent bonds formed between W atoms and their neighboring atoms.