摘要
由于受到严苛的服役环境和中子辐照的影响,核动力装置用奥氏体不锈钢作为结构材料应用时对力学性能要求较高,因此对于奥氏体不锈钢力学性能的预测很值得关注和研究。将机器学习算法应用于材料信息学并对机器学习的方法和原理作了简要说明,重点介绍了基于奥氏体不锈钢力学性能数据库,以奥氏体不锈钢力学性能预测为应用实例建立了机器学习模型和系统平台,最后通过预测值与真实值的对比验证对模型进行了评估。研究结果表明,构建的相关模型可以对奥氏体不锈钢的抗拉强度和屈服强度进行有效预测,R^(2)均在0.90以上。对现阶段机器学习在性能预测和材料研发领域急需解决的问题进行了探讨,并对其未来的发展方向进行了展望。
Due to the harsh service environment and the influence of neutron irradiation, the application of austenitic stainless steel in nuclear power plants as a structural material requires high mechanical properties. Therefore, the prediction of the mechanical properties of austenitic stainless steel is worthy of attention and research. Machine learning algorithm was applied to material informatics and the methods and principles of machine learning were briefly described. At the same time, the machine learning model and system platform based on the prediction of mechanical properties of austenitic stainless steels were introduced. Finally, the model was evaluated by comparing the predicted value with the real value. The research results show that the constructed model can predict the tensile strength and yield strength of austenitic stainless steel, and R^(2) is above 0.90. At last, the problems that need to be solved urgently in the field of performance prediction and material research and development of machine learning at this stage were discussed, and its future development direction was looked forward.
作者
王卓
朱虹
许斌
宋丹戎
王留兵
张宏亮
WANG Zhuo;ZHU Hong;XU Bin;SONG Danrong;WANG Liubing;ZHANG Hongliang(Light Alloy Research Institute,Central South University,Changsha 410083,Hunan,China;MatAi,Chengdu 610041,Sichuan,China;Science and Technology on Reactor System Design Technology Laboratory,Chengdu 610213,Sichuan,China;Nuclear Power Institute of China,Chengdu 610213,Sichuan,China)
出处
《钢铁研究学报》
CAS
CSCD
北大核心
2023年第2期201-209,共9页
Journal of Iron and Steel Research
基金
核反应堆系统设计技术重点实验室资助项目(LRSDT2018211)
四川省科技计划资助项目(2019ZDZX0001)。
关键词
核电站
奥氏体不锈钢
机器学习
性能预测
数据库
nuclear power plant
austenitic stainless steel
machine learning
performance prediction
database
