基于深度神经网络的ODS合金辐照硬化预测

Radiation Hardening Prediction for ODS Alloy Based on Deep Neural Network

  • 摘要: 氧化物弥散强化(ODS)合金作为第4代先进堆结构材料和聚变堆第一壁结构材料的候选材料,其抗辐照性能仍是制约其在快堆和聚变堆领域应用的关键技术难题。本文通过收集ODS合金的成分、固化和热处理工艺、辐照条件、测试条件(包括温度等)及屈服强度等数据约570条,并对数据进行清洗及重要属性的筛选,采用机器学习中反向传播的深度神经网络方法,尝试构建了Cr、Y2O3等关键成分与ODS合金中子辐照硬化的关联性,获得针对ODS合金辐照硬化的性能预测。结果表明:Cr含量约为6%、Y2O3添加量约为0.2%时,ODS合金的辐照硬化程度降低,同时Ti的添加有利于ODS合金辐照硬化程度的降低,而添加Al则会加剧其辐照硬化。据此,后续可得出一定辐照条件下,辐照硬化程度更低的ODS合金成分设计方案。

     

    Abstract: As the candidate structural material for the fourth generation advanced reactor and the first wall for fusion reactor, the irradiation resistance of oxide dispersion strengthened (ODS) alloy is still a key problem restricting its application in fast reactor and fusion reactor. In this paper, 570 groups of data containing composition, sintering and heat treatment process, irradiation condition, test condition and yield strength were collected. The data above were cleaned and screened before model training. By using the deep neural network with back propagation, the relationship between Cr, Y2O3 and other key component and neutron irradiation hardening of ODS alloy was established, and the irradiation hardening prediction of ODS alloy was obtained. The results show that the irradiation hardening of ODS alloy decreases when the content of Cr is about 6% and the content of Y2O3 is about 0.2%. Besides, the addition of Ti is beneficial to reduce the irradiation hardening of ODS alloy, while the addition of Al will aggravate the irradiation hardening. Based on these, the composition design of ODS alloy with less irradiation hardening under certain irradiation condition can be obtained further.

     

/

返回文章
返回