基于奇异值分解的γ能谱弱峰识别  

Weak peak identification of gamma spectrum based on singular value decomposition

作　　者：陈锋 周建斌[1] 刘易 CHEN Feng;ZHOU Jianbin;LIU Yi(College of Nuclear Technology and Automation Engineering,Chengdu University of Technology,Chengdu 610059,China)

机构地区：[1]成都理工大学核技术与自动化工程学院,成都610059

出　　处：《核技术》2024年第9期93-101,共9页Nuclear Techniques

基　　金：国家自然科学基金(No.12075038);四川省自然科学基金(No.2024NSFSC0423)资助。

摘　　要：针对低放射性核素样品的γ能谱分析中的弱峰识别问题,提出了一种基于奇异值分解的γ能谱寻峰新方法。该方法通过改进矩阵的构造方式,将γ能谱升维为双向循环矩阵后进行奇异值分解,选择第二个奇异值进行矩阵重构,并根据重构后的矩阵寻峰。以放射源152Eu的γ能谱为实验对象,与一阶导数寻峰、对称零面积寻峰和其他矩阵的奇异值分解寻峰方法进行了对比。实验结果表明:双向循环矩阵奇异值分解寻峰方法,在检测弱峰时拥有更高的查全率、查准率和F1值,分别达到了100%、87%和0.94,优化了弱峰检测的效果,为寻峰方法提供更多的选择。[Background]When performing gamma-ray spectroscopy analysis of samples with low levels of radioactive nuclide content,the weak peaks are difficult to be identified.[Purpose]This study aims to propose a new method for identifying peaks inγspectra by utilizing singular value decomposition(SVD)to improve the detection efficiency of weak peaks.[Methods]Firstly,the matrix construction of spectrum data was improved by transforming theγspectrum into a two-way cyclic matrix,and singular value decomposition of matrix was performed to get singular values and singular vectors.Then,the second singular value was selected to reconstruct the matrix and perform peak finding.Finally,theγspectrum of the radioactive source 152Eu was used as the experimental object,the peak finding performance of proposed method was compared with that of first-order derivative peak finding,symmetric zero-area peak finding,and singular value decomposition peak finding.[Results]Comparison result show that the bidirectional circular matrix SVD peaking method has higher recall rate,precision rate,and F1 value,achieving 100%,87%and 0.94,respectively.[Conclusions]The approach of this study can optimize weak peak detection and offer additional options for peak finding methods.

关 键 词：弱峰检测 奇异值分解 双向循环矩阵 

分 类 号：TL99[核科学技术—核技术及应用]