基于RBF人工神经网络的γ能谱分析  

γ- Spectra Analysis Based on RBF Artificial Neural Network

作　　者：王崇杰[1] 贾慧慧[1] 冯琳懿 刘硕[1] 刘媛媛[1] 张倩妮[1] 张博超[1] 

机构地区：[1]辽宁师范大学物理与电子技术学院,大连116029

出　　处：《核电子学与探测技术》2016年第1期56-59,共4页Nuclear Electronics & Detection Technology

基　　金：辽宁省教育厅科学技术研究项目(No.L2013410)资助

摘　　要：提出了基于人工神经网络的γ能谱定量分析方法。通过对标准源γ能谱的线性组合,建立了γ能谱分析网络训练样本集,并对RBF人工神经网络进行了训练。对已知标准点源不同测量时间的组合γ能谱和标准环境样品γ能谱进行了仿真分析,分析结果的相对误差分别小于3%和小于8%。研究表明,基于RBF人工神经网络的γ能谱分析方法是有效可行的。A method of analyzing γ- spectra based on RBF ANN was introduced. The training sample sets were established by linearly combining the spectra of the standard radioactive sources that contain a single nuclide composition respectively,and the RBF ANN was trained with these sample sets. The combination spectra by different collecting time of the standard point sources and those of a standard environmental sample were analyzed.The deviation was less than 3% and 8% respectively. The results show that the method of analyzing γ- spectra based on RBF ANN is effective and feasible.

关 键 词：人工神经网络 RBF网络 γ能谱分析 线性时不变系统 线性叠加原理 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]