期刊文献+
共找到2篇文章
< 1 >
每页显示 20 50 100
Highly oriented NdFeB nanocrystalline magnets from partially recombined compacts with ultrafine grain size by reactive deformation under low pressure 被引量:5
1
作者 余云萍 李军 +3 位作者 刘颖 王仁全 郑青 连利仙 《Journal of Rare Earths》 SCIE EI CAS CSCD 2015年第12期1298-1302,共5页
The partially recombined compacts with ultrafine grain size were taken advantage of preparing anisotropic nanocrystalline magnets with full density and homogenous microstructure and texture by reactive deformation und... The partially recombined compacts with ultrafine grain size were taken advantage of preparing anisotropic nanocrystalline magnets with full density and homogenous microstructure and texture by reactive deformation under low pressure. Because of the ul- trafine grain size of the precursors, the partially recombined phases could quickly achieve recombination. The results suggested that the newly recombined Nd2Fe14B grains with fme grain size could undergo deformation immediately during the desorp- tion-recombination reaction, and then an obvious anisotropy and uniform alignment would be obtained. The magnetic properties, (BH)max=214 kJ/m3, Br= 1.26 T, Hcj=463 kA/m, were obtained after being treated for 5 min at 820 ℃ in high vacuum under low pres- sure less than 26 MPa. Microstructures of the magnets were analyzed using X-ray diffraction (XRD), scanning electron microscopy (SEM) and transmission electron microscopy (TEM) respectively. Magnetic measurements were carried out using a vibrating sample magnetometer (VSM) with the maximum field of 2.88 T. Accurate phase contents were measured by a Mossbauer spectrometer. 展开更多
关键词 ndzfe14b NANOCRYSTALLINE HDDR deformation anisotropy rare earths
原文传递
用Bayesian正则化BP神经网络预测稀土永磁体性能
2
作者 王向中 查五生 +1 位作者 刘锦云 储林华 《电子元件与材料》 CAS CSCD 北大核心 2009年第8期75-77,85,共4页
针对一般BP神经网络泛化能力差,在Bayesian正则化BP神经网络的基础上,运用加权检验、"表决网"等方法的思路训练网络,并通过主成分分析方法对输入数据进行降维,建立了磁粉制备工艺(淬速度和晶化退火温度)、合金成分与磁性能之... 针对一般BP神经网络泛化能力差,在Bayesian正则化BP神经网络的基础上,运用加权检验、"表决网"等方法的思路训练网络,并通过主成分分析方法对输入数据进行降维,建立了磁粉制备工艺(淬速度和晶化退火温度)、合金成分与磁性能之间的BPNN(back propagation network)预测模型。结果表明:该模型泛化能力较高,预测的Br相对误差在2%左右、Hcj和(BH)max都在5%以内,且每次预测的相对误差平均值波动不超过1%。 展开更多
关键词 纳米晶复相(Nd2Fe14b/α-Fe)永磁体 主成分分析 bAYESIAN 正则化bP神经网络 泛化
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部