Machine Learning technique for isotopic determination of radioisotopes using HPGe -ray spectra

Aug 5, 2026

Upload by Patrick

HPGe 射线光谱测量是一种可用于放射性核素鉴别与同位素定量评估的定量、非破坏技术。传统同位素/富集度确定方法往往伴随统计与系统不确定性，并且通常需要多步预处理；同时在实验室条件下、屏蔽条件有限的情况下验证较少。本文在核威胁检测与应急响应场景中，引入多种机器学习回归算法作为替代方案，减少分析流程步骤，从而降低系统不确定性来源，并在应急响应应用中体现与传统方法相当的性能。

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages17

Reading Time43 min

AccessFree to download

Abstract

Introduction

Traditional methods and motivation

Non-destructive assay and automated algorithms

Organization of Paper

Background

Traditional Methods

arXiv:2301.01415v1 [[physics.data-an](physics.data-an)] 4 Jan 2023 Machine Learning technique for isotopic determination of radioisotopes using HPGe 􀀍-ray spectra Ajeeta Khatiwada, Marc Klasky, Marcie Lombardi, Jason Matheny, Arvind Mohan Los Alamos National Laboratory, Los Alamos, NM 87545, USA Abstract 􀀍-ray spectroscopy is a quantitative, non-destructive technique that may be utilized for the identiﬁcation and quantitative isotopic estimation of radionuclides. Traditional methods of isotopic determination have various challenges that contribute to statistical and systematic uncertainties in the estimated isotopics. Furthermore, these methods typically require numerous pre-processing steps, and have only been rigorously tested in laboratory settings with limited shielding. In this work, we examine the application of a number of machine learning based regression algorithms as alternatives to conventional approaches for analyzing 􀀍-ray spectroscopy data in the Emergency Response arena. This approach not only eliminates many steps in the analysis procedure, and therefore o􀀋ers potential to reduce this source of systematic uncertainty, but is also shown to o􀀋er comparable performance to conventional approaches in the Emergency Response Application. Keywords: Radionuclides, 􀀍-ray spectroscopy, Isotopic determination, enrichment determination, Machine Learning, Nuclear safeguards, Nuclear Threat Detection 1. Introduction The identiﬁcation and quantitative determination of the isotopic content of samples/objects potentially containing uranium and/or plutonium is of paramount importance to the nuclear materials safeguards, arms control veriﬁcation, nuclear security, Emergency Response (ER), as well in nuclear remediation arenas [1, 2, 3, 4, 5, 6, 7, 8, 9, 10] . Conventional methods for determining the isotopics/enrichment using 􀀍-ray spectroscopy require many time consuming steps [1. photo-peak identi](1. photo-peak identi)ﬁcation, 2. background and continuum subtraction, 3. feature extraction, 4. estimation of the relative e􀀎ciency curve, and 5. matching of the extracted features with those of known nuclides to estimate the fraction of isotopes [11] . In many of these application areas, it is imperative to rapidly determine the isotopic fractions using remote detection techniques. These constraints necessitate the 􀀃 Corresponding Author Email address: [ajeeta@lanl.gov](ajeeta@lanl.gov) (Ajeeta Khatiwada) use of non-destructive assay methods (NDA) and accompanying automated algorithms to perform quantitative analysis. In some applications, details regarding the physical arrangement of the nuclear materials cannot be revealed due to security concerns, [e.g. in](e.g. in)[ ](e.g. in)[treaty veri](treaty veri)ﬁ[cation activities](cation activities), [or are unknown e.g](or are unknown e.g). in nu clear security and ER activities in which the shielding and other aspects of the physical conﬁguration are unknown. In this work, we examine the ability of numerous machine learning (ML) techniques to address the automated identiﬁcation and quantiﬁcation of uranium and plutonium isotopics for ER applications. 2. Organization of Paper In this work, we investigate the application of a variety of machine learning algorithms to perform uranium and plutonium isotopic estimation for Emergency Response applications. Before discussing the ML algorithms utilized in these investigations, we present a review of both the traditional as well as the ML methods to perform quantitative isotopic identiﬁcation in Section 3 . The machine algorithms utilized in this investigation are presented in Section 4 . In Section 5, the generation of ML training data is discussed along with an Preprint submitted to Elsevier January 5, 2023 investigation of the accuracy of these simulations to emulate experimental data. Details of the pre-processing of the spectral data including background, continuum subtraction, and feature extr

FAQs

HPGe 射线光谱在同位素确定中用于什么目的？

它用于对放射性核素进行定量、非破坏的鉴别，并估计同位素含量（以及富集度）。

传统同位素确定方法的主要挑战是什么？

传统方法步骤多，包含多种预处理环节（如峰识别、背景扣除、特征提取、效率曲线估计与特征匹配），从而引入统计与系统不确定性。

本文如何利用机器学习改进应急响应场景下的分析？

通过使用机器学习回归算法替代部分传统流程，减少分析步骤，并在应急响应应用中与传统方法达到可比性能。

Aug 5, 2026

Upload by Patrick