基于人工神经网络的RPV材料辐照脆化预测模型研究  

Research on Prediction Model of Irradiation Embrittlement of RPV Materials Based on Artificial Neural Network

作　　者：康靓 孙凯[2] 米晓希 吴璐[2] 毛建军 张烁 雷阳[2] 潘荣剑 汤爱涛[1] Kang Jing;Sun Kai;Mi Xiaoxi;Wu Lu;Mao Jianjun;Zhang Shuo;Lei Yang;Pan Rongjian;Tang Aitao(School of Material Science and Technology,Chongqing University,Chongqing,400044,China;Nuclear Power Institute of China,Chengdu,610213,China)

机构地区：[1]重庆大学材料科学与工程学院,重庆400044 [2]中国核动力研究设计院,成都610213

出　　处：《核动力工程》2020年第6期92-95,共4页Nuclear Power Engineering

摘　　要：在分析一定量随站测试样品的基础上,构建了具有较高精度的反应堆压力容器(RPV)材料韧脆转变温度(DBTT)预测的人工神经网络模型,并利用模型研究了中子注量和中子注量率对RPV材料DBTT的影响。结果表明,材料DBTT随着中子注量增加出现先线性上升,然后平缓上升,最后饱和的趋势,而中子注量率对RPV材料辐照脆化的影响不明显。Based on the analysis of a certain amount of on-site test samples,this paper constructs a high-precision artificial neural network model for the ductile-brittle transition temperature prediction of RPV materials.Then we use the model to explore the influence of neutron fluence and neutron fluence rate parameters on the ductile-brittle transition temperature of RPV materials.It is found that the ductile-brittle transition temperature increases linearly with the increasing of neutron fluence,and then rises slowly and finally saturates.The effect of neutron flux rate on the embrittlement of RPV materials is not obvious.

关 键 词：反应堆压力容器(RPV) 误差逆向传播的神经网络模型(BP-ANN) 辐照脆化 韧脆转变温度(DBTT) 

分 类 号：TL351.6[核科学技术—核技术及应用] TP183[自动化与计算机技术—控制理论与控制工程]