机器学习辅助高熵合金相结构预测

Phase Prediction of High Entropy Alloys Aided by Machine Learning

高熵合金由于其形成独特显微组织的固溶体、金属间化合物和非晶相而具有更好的物理化学性能.因此,高熵合金中的相预测是合金设计的第一步.采用机器学习算法中的支持向量机、随机森林和决策树3种模型对高熵合金的相位分类进行预测,通过网格搜索方法优化模型,并对模型进行交叉验证和性能评估.结果表明:随机森林的预测能力最佳,达到0.93的预测精度,且该模型对高熵合金固溶体相的分类效果最好,最后采用随机森林模型预测Ti Zr Nb Mo系难熔高熵合金的生成相,其预测生成相与实验结果一致.由此可见,机器学习技术对未来高熵合金的设计有很大的帮助.

High entropy alloys have better physical and chemical properties due to the formation of solid solution( SS),intermetallic compound( IM) and amorphous phase( AM) with unique microstructure.Therefore,the prediction of phase in high entropy alloys is the first step in alloy design. In this paper,support vector machine( SVM),random forest( RF) and decision tree( DT) models in machine learning( ML) algorithm were used to predict phase classification in high entropy alloys. The mesh search method was used to optimize the model,and the cross-validation and performance evaluation of the model were carried out. The results showed that: random forest had the best prediction ability,reaching a prediction accuracy of 0. 93 and this model had the best classification effect on SS of high entropy alloys. Finally,a trained RF model was used to predict the phase formation of Ti Zr NbMo alloys. The predicted phase formation was consistent with the experimental results. Therefore,machine learning technology is of great help to the design of future high entropy alloys.