基于PCA-FCN混合模型的NaI(Tl)伽马能谱核素识别技术研究

辐射防护 ›› 2025, Vol. 45 ›› Issue (4): 327-336.

基于PCA-FCN混合模型的NaI(Tl)伽马能谱核素识别技术研究

刘鑫^1,2, 赵日^1,2, 谭俊¹, 王茂林¹, 黄健¹, 张静^1,2, 梁润成^1,2, 刘兆行^1,2, 石忠焱^1,2, 王佳¹, 令狐仁静¹, 刘立业¹

1.中国辐射防护研究院,太原 030006;
2.核药研发转化与精准防护山西省重点实验室,太原 030006

收稿日期:2024-10-24 发布日期:2025-07-28
通讯作者: 刘立业。E-mail:liuliye@cirp.org.cn
作者简介:刘鑫(1994—),男,2019年毕业于兰州大学原子核物理专业,获学士学位,2022年毕业于兰州大学粒子物理与原子核物理专业,获硕士学位,助理研究员。E-mail:xonleo@163.com

Research on NaI(Tl) gamma spectrum radionuclide identification technology based on hybrid PCA-FCN model

LIU Xin^1,2, ZHAO Ri^1,2, TAN Jun¹, WANG Maolin¹, HUANG Jian¹, ZHANG Jing^1,2, LIANG Runcheng^1,2, LIU Zhaoxing^1,2, SHI Zhongyan^1,2, WANG Jia¹, LINGHU Renjing¹, LIU Liye¹

1. China Institute for Radiation Protection,Taiyuan 030006;
2. Shanxi Provincial Key Laboratory for Translational Nuclear Medicine and Precision Protection, Taiyuan 030006

Received:2024-10-24 Published:2025-07-28

摘要/Abstract

摘要： NaI(Tl)探测器能量分辨率较差使得基于其获取的伽马能谱进行准确的核素识别较为困难,为提高识别准确率,综合已有研究方法和模型的优缺点,提出了PCA-FCN混合识别模型,并基于随机化策略通过实验测量和蒙特卡罗模拟构建了有较强通用性的γ能谱数据集,利用数据集对模型进行训练并开展实验测量能谱验证。结果表明,PCA-FCN模型的核素识别A_P、F₁性能因子达到0.982 3和0.980 1,显著优于PCA模型、FCN模型和传统全能峰分析法,而且在不同能谱复杂度、不同能谱统计涨落下仍能保持识别准确性。该结论显示了PCA-FCN模型和随机化样本生成策略在未来放射性定量测量应用的潜力。

关键词: NaI(Tl)探测器, 核素识别, 伽马能谱, 主成分分析, 全连接网络

Abstract: The poor energy resolution of NaI(Tl) detector makes it difficult to accurately identify radionuclides based on the acquired gamma spectrum. In order to improve the identification accuracy, this paper proposed a hybrid PCA-FCN identification model based on the advantages and disadvantages of existing research methods and models. Based on the randomization strategy, a robust gamma energy spectrum datasets was constructed through empirical measurements and Monte Carlo simulations. The datasets was utilized to train the model and conduct validation experiments. Results indicate that the PCA-FCN model achieved radionuclide identification performance factors of 0.982 3 for average precision (A_P) and 0.980 1 for the F₁ score, markedly surpassing the performance of PCA models, FCN models, and traditional full-energy peak analysis methods. The identification accuracy can still be maintained under varying spectral complexities and statistical fluctuations. This conclusion shows the potential of the PCA-FCN model and the random sample generation strategy for future applications in quantitative measurement of radioactivity.

Key words: NaI(Tl) detector, radionuclide identification, gamma spectrum, principal component analysis, fully connected network

中图分类号:

刘鑫, 赵日, 谭俊, 王茂林, 黄健, 张静, 梁润成, 刘兆行, 石忠焱, 王佳, 令狐仁静, 刘立业. 基于PCA-FCN混合模型的NaI(Tl)伽马能谱核素识别技术研究[J]. 辐射防护, 2025, 45(4): 327-336.

LIU Xin, ZHAO Ri, TAN Jun, WANG Maolin, HUANG Jian, ZHANG Jing, LIANG Runcheng, LIU Zhaoxing, SHI Zhongyan, WANG Jia, LINGHU Renjing, LIU Liye. Research on NaI(Tl) gamma spectrum radionuclide identification technology based on hybrid PCA-FCN model[J]. RADIATION PROTECTION, 2025, 45(4): 327-336.

参考文献

[1] 高妍,曾国强,顾民,等.核素识别仪中的自适应变换宽度SNIP算法开发[J].核技术,2019,42(6): 30-34.
GAO Yan, ZENG Guoqiang, GU Min, et al. Development of adaptive width SNIP algorithm in nuclides identification instrument[J]. Nuclear Techniques, 2019, 42(6): 30-34.
[2] 赵日,刘立业,李君利.基于递推关系式的γ能谱全能峰基底扣减方法[J].原子能科学技术,2018, 52(6):1085-1091.
ZHAO Ri, LIU Liye, LI Junli. Continuum subtraction method under full-energy peak in γ spectrum based on recursion formula[J]. Atomic Energy Science and Technology, 2018, 52(6): 1085-1091.
[3] ZHU Menghua, LIU Lianggang, QI Dongxu, et al. Least square fitting of low resolution gamma ray spectra with cubic B-spline basis functions[J]. Chinese Physics C, 2009, 33(1): 24-30.
[4] BI Haijie, ZHANG Yingying, WU Bingwei, et al. Study on reconstruction and analytical method of seawater radioactive gamma spectrum[J]. Applied Radiation and Isotopes: Including Data, Instrumentation and Methods for Use in Agriculture, Industry and Medicine, 2023, 198: 110853.
[5] 赵日,刘立业,曹勤剑.基于能谱重建技术的全身计数器γ能谱分析[J].原子能科学技术,2019,53(8): 1495-1501.
ZHAO Ri, LIU Liye, CAO Qinjian. γ spectrum analysis based on reconstruction technology for whole body counter[J]. Atomic Energy Science and Technology, 2019, 53(8): 1495-1501.
[6] 霍勇刚,安洪振,吴岳雷,等.一种基于模式识别的放射性核素快速识别方法研究[J].核电子学与探测技术,2014,34(1):51-53.
HUO Yonggang, AN Hongzhen, WU Yuelei, et al. Fast radioactive nuclide recognition method study based on pattern recognition[J]. Nuclear Electronics & Detection Technology, 2014, 34(1): 51-53.
[7] 张江梅,任俊松,李培培,等.基于支持向量机的复杂核素能谱识别[J].核电子学与探测技术,2016, 36(8):856-861.
ZHANG Jiangmei, REN Junsong, LI Peipei, et al. Complex radioactive nuclide identification method based on support vector machine[J]. Nuclear Electronics & Detection Technology, 2016, 36(8): 856-861.
[8] LI Xiaozhe, ZHANG Qingxian, TAN Heyi, et al. Fast nuclide identification based on a sequential Bayesian method[J]. Nuclear Science and Techniques, 2021, 32(12): 143.
[9] QI Sheng, WANG Shanqiang, CHEN Ye, et al. Radionuclide identification method for NaI low-count gamma-ray spectra using artificial neural network[J]. Nuclear Engineering and Technology, 2022, 54(1): 269-274.
[10] Kim J, Park K, Cho G. Multi-radioisotope identification algorithm using an artificial neural network for plastic gamma spectra[J]. Applied Radiation and Isotopes, 2019, 147: 83-90.
[11] Kamuda M, Stinnett J, Sullivan C J. Automated isotope identification algorithm using artificial neural networks[J]. IEEE Transactions on Nuclear Science, 2017, 64(7): 1858-1864.
[12] CHEN Liang, WEI Yixiang. Nuclide identification algorithm based on K–L transform and neural networks[J]. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 2009, 598(2): 450-453.
[13] Mathew J, Kshirsagar R, Abidin D Z, et al. A comparison of machine learning methods to classify radioactive elements using prompt-gamma-ray neutron activation data[J]. Scientific Reports, 2023, 13(1): 9948.
[14] Al-Nima R R O, Dlay S S, Woo W L, et al. A novel biometric approach to generate ROC curve from the Probabilistic Neural Network[C]//2016 24th Signal Processing and Communication Application Conference (SIU):IEEE, 2016: 141-144.