基于机器学习的RAFM钢中子辐照脆化预测模型研究  

Research on Prediction Model of Neutron Irradiation Embrittlement of RAFM Steels Based on Machine Learning

作　　者：李孝晨 丁文艺 朱霄汉 郑明杰 LI Xiaochen;DING Wenyi;ZHU Xiaohan;ZHENG Mingjie(Hefei Institutes of Physical Science,Chinese Academy of Sciences,Hefei 230031,China;University of Science and Technology of China,Hefei 230026,China)

机构地区：[1]中国科学院合肥物质科学研究院,合肥230031 [2]中国科学技术大学,合肥230026

出　　处：《材料导报》2023年第1期147-153,共7页Materials Reports

基　　金：国家重点研发计划项目(2018YFE0307104);国家自然科学基金(11632001);中国科学院合肥物质科学研究院院长基金国际合作探索项目(2021YZGH05);中国科学院特别交流计划A类(E2AAAI13)。

摘　　要：构建低活化铁素体/马氏体(RAFM)钢的中子辐照脆化预测模型对聚变反应堆的安全运行和优化设计新型RAFM钢具有十分重要的意义。本研究基于收集的RAFM钢中子辐照数据集,采用相关性筛选、递归消除方法识别出影响RAFM钢中子辐照条件下韧脆转变温度(DBTT)的关键特征变量。利用筛选的关键特征变量,构建了具有良好预测能力的RAFM钢中子辐照DBTT预测模型。为进一步实现中子辐照条件下韧脆转变温度变化(ΔDBTT)的预测,首先构建了RAFM钢未辐照DBTT预测模型,然后将辐照前后DBTT预测模型相结合构建了RAFM钢中子辐照ΔDBTT预测模型。通过将模型预测的ΔDBTT与文献收集的数据进行对比发现,该模型具备较好的预测能力。Building the neutron irradiation embrittlement prediction model of reduced activation ferritic/martensitic(RAFM)steels is of great significance for the safe operation of the fusion reactor and the optimal design of new RAFM steels.In the present work,based on the collected neutron irradiation dataset of RAFM steels,the key features affecting the ductile-brittle transition temperature(DBTT)of RAFM steels under neutron irradiation are identified by correlation screening and recursive elimination methods.Using the selected key features,the prediction model for DBTT of neutron-irradiated RAFM steels with good prediction ability is constructed.In order to further predict the ductile-brittle transition temperature shift(ΔDBTT)under neutron irradiation,the prediction model for DBTT of un-irradiated RAFM steels is constructed.The prediction model forΔDBTT of neutron-irradiated RAFM steels is constructed by combining the prediction models before and after irradiation.By comparing theΔDBTT predicted by the model with the data collected from the related experimental literatures,the results indicate that this prediction model has high precision and reliability.

关 键 词：机器学习 RAFM钢 辐照脆化 韧脆转变温度 

分 类 号：TG113.25[金属学及工艺—物理冶金]