期刊文献+

CNN和BPNN核素识别算法的对比 被引量:1

Comparison of Nuclide Identification Algorithms Based on CNN and BPNN
下载PDF
导出
摘要 研究目的为比较卷积神经网络与反向传播神经网络在核素识别中的效果。将能谱数据变换为灰度图像,作为卷积神经网络的输入;变换后的图像经过奇异值分解提取特征,作为反向传播神经网络的输入。两类网络进行训练后,对它们在核素识别中的性能进行评估、测试和对比。结果表明:利用深度学习的方法可以进行核素识别。两种方法中,卷积神经网络较于反向传播神经网络性能更优;奇异值分解的方法能够提高反向传播神经网络的识别效果。 The aim was to compare the performances of two nuclide identification algorithms based on deep learning.Energy spectrum data were transformed into images as the input of Convolutional Neural Network.Singular Value Decomposition was used to extract the features of the transformed image as the input of the Back Propagation Neural Network.The two kinds of networks were trained and their performances in nuclide identification were compared and evaluated.The results showed that the deep learning method was feasible and efficient for nuclide identification.Singular Value Decomposition method could improve the recognition effect of Back Propagation Neural Network.The performance of the Convolutional Neural Network was better than that of Back Propagation Neural Network.
作者 吴泇俣 王世磊 胥建国 曹文田 WU Jia-yu;WANG Shi-lei;XU Jian-guo;CAO Wen-tian(Institute of Heavy Ion Physics,School of Physics,Peking University,Beijing 100871,China;Unit 92609,Beijing 100077,China)
出处 《核电子学与探测技术》 CAS 北大核心 2021年第6期966-973,共8页 Nuclear Electronics & Detection Technology
基金 国家重点研发计划(2019YFF01014405)资助。
关键词 核素识别 深度学习 卷积神经网络 反向传播神经网络 Nuclide Identification Deep Learning CNN BPNN
  • 相关文献

参考文献4

二级参考文献19

共引文献29

同被引文献5

引证文献1

二级引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部