基于KAN和MNv4的中子/伽马甄别方法

王雨龙; 翁秀峰; 刘霄; 陈翔; 盛亮; 刘彤

doi:10.11889/j.0253-3219.2025.hjs.48.240349

您当前的位置：

首页 >

文章列表页 >

基于KAN和MNv4的中子/伽马甄别方法

核电子学与仪器 | 更新时间：2025-06-16

- 基于KAN和MNv4的中子/伽马甄别方法
- Neutron and gamma discrimination method based on KAN and MNv4
- 核技术 2025年48卷第6期文章编号：060401
- 作者机构：
  
  1.上海交通大学核科学与工程学院上海 200240
  2.西北核技术研究所西安 710024
  3.强脉冲辐射模拟与效应全国重点实验室西安 710024
- 作者简介：
  
  王雨龙，男，2000年出生，2022年毕业于上海电力大学，现为硕士研究生，研究领域为空间粒子探测
  盛亮，E-mail：shengliang@nint.ac.cn
  刘彤，E-mail：tongliu@sjtu.edu.cn
- 基金信息：
  
  卓越青年基金(JQZQ021901);核技术研发科研项目(HNKF202314(48))
- DOI：10.11889/j.0253-3219.2025.hjs.48.240349
  中图分类号： TL812
- 收稿日期：2024-09-06，
  
  修回日期：2024-12-24，
  
  纸质出版日期：2025-06-15
- 稿件说明：
移动端阅览
王雨龙,翁秀峰,刘霄等.基于KAN和MNv4的中子/伽马甄别方法[J].核技术,2025,48(06):060401.

WANG Yulong,WENG Xiufeng,LIU Xiao,et al.Neutron and gamma discrimination method based on KAN and MNv4[J].NUCLEAR TECHNIQUES,2025,48(06):060401.
王雨龙,翁秀峰,刘霄等.基于KAN和MNv4的中子/伽马甄别方法[J].核技术,2025,48(06):060401. DOI： 10.11889/j.0253-3219.2025.hjs.48.240349. CSTR： 32193.14.hjs.CN31-1342/TL.2025.48.240349.

WANG Yulong,WENG Xiufeng,LIU Xiao,et al.Neutron and gamma discrimination method based on KAN and MNv4[J].NUCLEAR TECHNIQUES,2025,48(06):060401. DOI： 10.11889/j.0253-3219.2025.hjs.48.240349. CSTR： 32193.14.hjs.CN31-1342/TL.2025.48.240349.

摘要

实现对空间中子/伽马（n/γ）的精准甄别有利于空间辐射场的研究。传统的甄别方法受数据噪声影响较大，难以实现在星载探测器这样高噪声水平条件下进行n/γ甄别。为实现在不同噪声水平下n/γ的有效甄别，本研究提出了使用小波散射网络（Scattering Convolution Networks，SCN）结合MobileNetV4（MNv4）和Kolmogorov–Arnold Networks（KAN）模型的n/γ甄别的算法。研究选用Ø50 mm×50 mm的NaI（Tl）：

Li晶体与H6410光电倍增管对中国原子能科学研究院国防科技工业电离辐射一级计量站的2 500 keV与5 000 keV的中子束流进行数据采集，将2 500 keV按信号触发电平分为测试集2与训练集，5 000 keV作为测试集1。将数据集通过电荷比较法进行分类以后，使用小波散射网络进行去噪降维处理，使1 800个样本点降低为113个，然后通过MNv4或KAN模型进行粒子甄别。然后通过外加10%~50%噪声水平大小的噪声模拟不同噪声下的信号以研究算法抗噪性，最后将下降沿触发为-4 mV、-8 mV、-12 mV的不同信噪比信号进行甄别，验证算法的鲁棒性。实验表明，MNv4与KAN在两个测试集识别准确率为99.7%与98.6%，相较于比较法能够将品质因子（Figure of Merit，FOM）从1.82提高到4.07与4.40。实验表明，在15%噪声水平内，MNv4和KAN对2 500 keV与5 000 keV能量的中子的甄别准确率为99.7%，在40%噪声水平时有着大于99%的甄别准确率，在信噪比（Signal-to-Noise Ratio

，SNR）为9.13 dB时具有99%以上甄别准确率。结合小波散射网络与MNv4或KAN的深度学习算法具有降维与抗噪的能力，相较于电荷比较法，能够提高甄别的品质因子，同时还能实现高噪声水平下的中子/伽马甄别，具有良好的鲁棒性。

Abstract

Background

Achieving precise identification of neutrons/gammas in space is beneficial for studying space radiation fields. Traditional identification methods are significantly affected by data noise

making it challenging to distinguish neutrons/gammas under the high-noise conditions of spaceborne detectors.

Purpose

This study aims to propose a deep learning algorithm that combines Scattering Convolution Networks with MobileNetV4 (MNv4) and Kolmogorov–Arnold Networks (KAN) to improve neutron/gamma discrimination performance and the noise robustness of this method.

Methods

Firstly

a Ø50 mm×50 mm NaI(Tl):

Li crystals and H6410 photomultiplier tubes were employed to collect data on neutron beams of 2 500 keV and 5 000 keV from the National Defense Technology Industrial Ionizing Radiation Calibration Station at China Institute of Atomic Energy (CIAE). Then

the 2 500 keV data were divided into test set 2 and training set

while the 5 000 keV data served as test set 1. After classifying the dataset using the charge comparison method

the scattering convolution networks (SCN) was used for denoising and dimensionality reduction

reducing 1 800 sample points to 113. Particle identification was then performed using MNv4 or KAN model. Finally

simulated signals with noise levels of 10%~50% were applied to investigating the algorithm's noise resistance and tested different signal-to-noise ratios with falling edge trigger leves of -4 mV

-8 mV

and -12 mV to verify the algorithm's robustness.

Results

The experiment results show that MNv4 and KAN achieve recognition accuracies of 99.7% and 98.6% on the two test sets

respectively

compared to the comparison method

which improves the Figure of Merit (FOM) from 1.82 to 4.07 and 4.40. At a 15% noise level

MNv4 and KAN achieve neutron identification accuracies of 99.7% for 2 500 keV and 5 000 keV energies whilst the identification accuracies of above two algorithms are maintained above 99% with a signal-to-noise ratio (SNR) of 9.13 dB at a 40% noise level.

Conclusions

Combining SCN with deep learning algorithms like MNv4 or KAN provides dimensionality reduction and noise resistance. Compared to the charge comparison method

this approach improves the Figure of Merit (FOM) and enables neutron/gamma discrimination even at high noise levels

demonstrating good robustness.

关键词

Keywords

references

Kegel G H R , DeSimone D . Facilities for gamma and neutron irradiations [J ] . Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , 1999 , 426 ( 1 ): 61 – 67 . DOI: 10.1016/s0168-9002(98)01470-3 http://dx.doi.org/10.1016/s0168-9002(98)01470-3 .

杨岐 , 卜永熙 , 李达忠 , 等 . 西安脉冲反应堆 [J ] . 核动力工程 , 2002 , 23 ( 6 ): 1 – 7 . DOI: 10.3969/j.issn.0258-0926.2002.06.001 http://dx.doi.org/10.3969/j.issn.0258-0926.2002.06.001 .

YANG Qi , PU Yongxi , LI Dazhong , et al . Xi'an pulsed reactor [J ] . Nuclear Power Engineering , 2002 , 23 ( 6 ): 1 – 7 . DOI: 10.3969/j.issn.0258-0926.2002.06.001 http://dx.doi.org/10.3969/j.issn.0258-0926.2002.06.001 .

张闯 . 国际粒子加速器的前沿 [J ] . 物理 , 2008 , 37 ( 5 ): 289 – 297 . DOI: 10.3321/0j.issn:0379-4148.2008.05.001 http://dx.doi.org/10.3321/0j.issn:0379-4148.2008.05.001 .

ZHANG Chuang . Frontiers of particle accelerators in the world [J ] . Physics , 2008 , 37 ( 5 ): 289 – 297 . DOI: 10.3321/j.issn:0379-4148.2008.05.001 http://dx.doi.org/10.3321/j.issn:0379-4148.2008.05.001 .

李名加 , 范娟 , 章法强 , 等 . 稠密等离子体焦点装置研制 [J ] . 强激光与粒子束 , 2018 , 30 ( 11 ): 180230 . DOI: 10.11884/HPLPB201830.180230 http://dx.doi.org/10.11884/HPLPB201830.180230 .

LI Mingjia , FAN Juan , ZHANG Faqiang , et al . Development of dense plasma focus device [J ] . High Power Laser and Particle Beams , 2018 , 30 ( 11 ): 180230 . DOI: 10.11884/HPLPB201830.180230 http://dx.doi.org/10.11884/HPLPB201830.180230 .

Craxton R S , Anderson K S , Boehly T R , et al . Direct-drive inertial confinement fusion: a review [J ] . Physics of Plasmas , 2015 , 22 ( 11 ): 110501 . DOI: 10.1063/1.4934714 http://dx.doi.org/10.1063/1.4934714 .

陈延伟 . 重器铸梦: 中国散裂中子源 [J ] . 现代物理知识 , 2023 , 35 ( S1 ): 205 – 209 . DOI: 10.13405/j.cnki.xdwz.2023.s1.022 http://dx.doi.org/10.13405/j.cnki.xdwz.2023.s1.022 .

CHEN Yanwei . Heavy equipment forging dreams: China's spallation neutron source [J ] . Modern Physics , 2023 , 35 ( S1 ): 205 – 209 . DOI: 10.13405/j.cnki.xdwz.2023.s1.022 http://dx.doi.org/10.13405/j.cnki.xdwz.2023.s1.022 .

Wang Z J , Xue Y Y , Tang N , et al . Synergistic radiation effects in PPD CMOS image sensors induced by neutron displacement damage and gamma ionization damage [J ] . Sensors , 2024 , 24 ( 5 ): 1441 . DOI: 10.3390/s24051441 http://dx.doi.org/10.3390/s24051441 .

Kunitoku Y , Akiyama Y , Manabe Y , et al . Study on irradiation effect of insulating materials for fusion superconducting magnet: effect of low-temperature irradiation [J ] . IOP Conference Series: Materials Science and Engineering , 2022 , 1241 ( 1 ): 012004 . DOI: 10.1088/1757-899X/1241/1/012004 http://dx.doi.org/10.1088/1757-899X/1241/1/012004 .

Yang J , Luo X L , Liu G F , et al . Digital discrimination of neutrons and γ rays with organic scintillation detectors in an 8-bit sampling system using frequency gradient analysis [J ] . Chinese Physics C , 2012 , 36 ( 6 ): 544 – 551 . DOI: 10.1088/1674-1137/36/6/011 http://dx.doi.org/10.1088/1674-1137/36/6/011 .

Liu M Z , Liu B Q , Zuo Z , et al . Toward a fractal spectrum approach for neutron and gamma pulse shape discrimination [J ] . Chinese Physics C , 2016 , 40 ( 6 ): 066201 . DOI: 10.1088/1674-1137/40/6/066201 http://dx.doi.org/10.1088/1674-1137/40/6/066201 .

Griffiths J , Kleinegesse S , Saunders D , et al . Pulse shape discrimination and exploration of scintillation signals using convolutional neural networks [J ] . Machine Learning: Science and Technology , 2020 , 1 ( 4 ): 045022 . DOI: 10.1088/2632-2153/abb781 http://dx.doi.org/10.1088/2632-2153/abb781 .

Roy A , Vartsky D , Mor I , et al . Evaluation of the Constant Fraction Time-Over-Threshold (CF-TOT) method for neutron-gamma pulse shape discrimination [J ] . Journal of Instrumentation , 2022 , 17 ( 5 ): P05028 . DOI: 10.1088/1748-0221/17/05/P05028 http://dx.doi.org/10.1088/1748-0221/17/05/P05028 .

Ye H Q , Chen L , Xu X Y , et al . Fast FPGA algorithm for neutron-gamma discrimination [J ] . Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , 2022 , 1027 : 166256 . DOI: 10.1016/j.nima. 2021.166256 http://dx.doi.org/10.1016/j.nima.2021.166256 .

Liu S X , Zhang W , Zhang Z H , et al . Performance of real-time neutron/gamma discrimination methods [J ] . Nuclear Science and Techniques , 2023 , 34 ( 1 ): 8 . DOI: 10.1007/s41365-022-01160-5 http://dx.doi.org/10.1007/s41365-022-01160-5 .

Chen J L , Guo J H , Wang S , et al . Improving neutron-gamma discrimination in CLYC with Gaussian mixture model [J ] . Journal of Instrumentation , 2023 , 18 ( 9 ): P09006 . DOI: 10.1088/1748-0221/18/09/p09006 http://dx.doi.org/10.1088/1748-0221/18/09/p09006 .

Yu X Z , Zhu J J , Lin S , et al . Neutron-gamma discrimination based on the support vector machine method [J ] . Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , 2015 , 777 : 80 – 84 . DOI: 10.1016/j.nima.2014.12.087 http://dx.doi.org/10.1016/j.nima.2014.12.087 .

Han J F , Zhu J L , Wang Z H , et al . Pulse characteristics of CLYC and piled-up neutron-gamma discrimination using a convolutional neural network [J ] . Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , 2022 , 1028 : 166328 . DOI: 10.1016/j.nima. 2022.166328 http://dx.doi.org/10.1016/j.nima.2022.166328 .

Peng S , Hua Z H , Wu Q , et al . Piled-up neutron-gamma discrimination system for CLLB using convolutional neural network [J ] . Journal of Instrumentation , 2022 , 17 ( 8 ): T08001 . DOI: 10.1088/1748-0221/17/08/T08001 http://dx.doi.org/10.1088/1748-0221/17/08/T08001 .

胡万平 , 张贵宇 , 张云龙 , 等 . 基于KPCA-MPA-ELM的n/γ甄别方法研究 [J ] . 核技术 , 2024 , 47 ( 4 ): 040403 . DOI: 10.11889/j.0253-3219.2024.hjs.47.040403 http://dx.doi.org/10.11889/j.0253-3219.2024.hjs.47.040403 .

HU Wanping , ZHANG Guiyu , ZHANG Yunlong , et al . Neutron/gamma (n/γ) discrimination method based on KPCA-MPA-ELM [J ] . Nuclear Techniques , 2024 , 47 ( 4 ): 040403 . DOI: 10.11889/j.0253-3219.2024.hjs.47.040403 http://dx.doi.org/10.11889/j.0253-3219.2024.hjs.47.040403 .

Ding T M , Jiang Y H , Wang X X , et al . Study on neutron-gamma discrimination methods based on GMM-KNN and LabVIEW implementation [J ] . Nuclear Science and Techniques , 2024 , 35 ( 11 ): 194 . DOI: 10.1007/s41365-024-01545-8 http://dx.doi.org/10.1007/s41365-024-01545-8 .

Liu H R , Zuo Z , Li P , et al . Anti-noise performance of the pulse coupled neural network applied in discrimination of neutron and gamma-ray [J ] . Nuclear Science and Techniques , 2022 , 33 ( 6 ): 75 . DOI: 10.1007/s41365-022-01054-6 http://dx.doi.org/10.1007/s41365-022-01054-6 .

Julier S J , Uhlmann J K . New extension of the Kalman filter to nonlinear systems [C ] . Signal Processing, Sensor Fusion, and Target Recognition VI . Orlando, FL, USA. SPIE , 1997 : 182 – 193 . DOI: 10.1117/12.280797 http://dx.doi.org/10.1117/12.280797 .

Xiao F , Zhang Y G . A comparative study on thresholding methods in wavelet-based image denoising [J ] . Procedia Engineering , 2011 , 15 : 3998 – 4003 . DOI: 10.1016/j.proeng.2011.08.749 http://dx.doi.org/10.1016/j.proeng.2011.08.749 .

Ulyanov D , Vedaldi A , Lempitsky V . Deep image prior [J ] . International Journal of Computer Vision , 2020 , 128 ( 7 ): 1867 – 1888 . DOI: 10.1007/s11263-020-01303-4 http://dx.doi.org/10.1007/s11263-020-01303-4 .

Ho J , Jain A , Abbeel P . Denoising diffusion probabilistic models [EB/OL ] . 2020: 2006 . 11239 . https://arxiv.org/abs/2006.11239v2 https://arxiv.org/abs/2006.11239v2 .

Bruna J , Mallat S . Invariant scattering convolution networks [J ] . IEEE Transactions on Pattern Analysis and Machine Intelligence , 2013 , 35 ( 8 ): 1872 – 1886 . DOI: 10.1109/TPAMI.2012.230 http://dx.doi.org/10.1109/TPAMI.2012.230 .

Qin D F , Leichner C , Delakis M , et al . MobileNetV4: universal models fortheMobile ecosystem [C ] // Computer Vision – ECCV 2024 . Cham : Springer Nature Switzerland , 2025 : 78 – 96 . DOI: 10.1007/978-3-031-73661-2_5 http://dx.doi.org/10.1007/978-3-031-73661-2_5 .

Liu Z , Wang Y , Vaidya S , et al . KAN: Kolmogorov-arnold networks [C/OL ] // The Thirteenth International Conference on Learning Representations . 2025 . https://openreview.net/forum?id=Ozo7qJ5vZi https://openreview.net/forum?id=Ozo7qJ5vZi .

Li W , Tao M R , Li Y Y , et al . NaI:Tl, 6 Li-PVT composite scintillator toward neutron and gamma discrimination [J ] . Radiation Measurements , 2022 , 159 : 106882 . DOI: 10.1016/j.radmeas.2022.106882 http://dx.doi.org/10.1016/j.radmeas.2022.106882 .

Shen F Z , Fu Q B , Huang T C , et al . A compact dual gamma neutron detector based on NaI(Tl+Li) scintillator readout with SiPM [J ] . Crystals , 2022 , 12 ( 8 ): 1077 . DOI: 10.3390/cryst12081077 http://dx.doi.org/10.3390/cryst12081077 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

深度学习在高能核物理中的前沿进展

基于深度学习的准部分子模型研究核物质状态方程

轻量级神经网络模型在核脉冲参数预测中的应用研究

基于KNN方法的熔盐堆系统瞬态识别模型开发及分析

嵌入位置编码的多头注意力机制在堆积脉冲幅度估计中的应用