基于机器学习的高强度ODS合金成分设计  

Composition Design of High Strength ODS Alloy Based on Machine Learning

作　　者：白冰[1] 郑全[1] 任帅 张长义[1] 杨文[1] 胡长军[2] BAI Bing;ZHENG Quan;REN Shuai;ZHANG Changyi;YANG Wen;HU Changjun(Division of Reactor Engineering Technology Research,China Institute of Atomic Energy,Beijing 102413,China;University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]中国原子能科学研究院反应堆工程技术研究部,北京102413 [2]北京科技大学,北京100083

出　　处：《原子能科学技术》2020年第4期678-682,共5页Atomic Energy Science and Technology

基　　金：中核集团集中研发项目资助(FA18000120);国防科工局核材料创新中心基金资助项目(ICNMT-2018-YZ-01)。

摘　　要：针对200~300组氧化物弥散强化(ODS)合金成分、工艺及力学性能数据,尝试借助机器学习的方法,建立了ODS合金中关键成分与拉伸性能的关联性。研究结果发现,在Cr、Y2O3、W和Ti含量与ODS合金抗拉强度的变化趋势中,均存在对应着抗拉强度极值的最优值,而添加Al对抗拉强度的提升无明显作用。获得了几种抗拉强度优化的ODS合金关键成分配比,预测出的室温抗拉强度均在1400 MPa以上,这将为快堆结构材料用ODS合金优化设计提供技术支持,加速推进ODS合金的材料优化。Based on 200-300 groups data of compositions,processes and mechanical properties,the relationship between the key parameters and tensile property of oxide dispersion strengthened(ODS)alloy was established by machine learning.The results show that the optimum value corresponding to the maximum strength exists in the relationship between the content of Cr,Y2O3,W and Ti and the tensile strength of ODS alloy.The addition of Al has no obvious effect on the increase of tensile strength.Therefore,several optimum compositions of ODS alloy were obtained,and the predicted tensile strength at room temperature is above 1400 MPa.It can help to promote the optimization of ODS alloy as cladding material for fast reactor application.

关 键 词：ODS合金 拉伸性能 机器学习 材料优化 

分 类 号：TG407[金属学及工艺—焊接]