AI提升宇宙粒子扫描精度,模拟中更准确定位隐藏物体

由 Gaby Clark 编辑,Robert Egan 审校
本文已按照 Science X 的编辑流程与政策进行审核。编辑在确保内容可信度的同时,突出了以下特征:
同行评审出版物
添加至首选来源
μ 子轨迹的多次库仑散射,以及散射角。该图展示了一个潜在应用:检查卡车内部以探测隐藏或违禁材料。来源:IEEE Access (2026)。DOI: 10.1109/access.2025.3648859
塔尔图大学计算机科学研究所的研究人员正与爱沙尼亚科技公司 GScan 合作,开发基于 AI 的方法,利用借助宇宙粒子采集的测量数据来更精准地定位物体。该技术可用于评估桥梁及其他大型结构的状况,并识别潜在损伤。
在塔尔图大学计算机科学研究所,讲师 Kallol Roy 及其研究同事正在开发智能方法,用于分析由一种名为 μ 子断层扫描(muon tomography)的特殊扫描技术所采集的信息。目标是帮助计算机判断大型物体内部隐藏着什么、这些物体位于何处,并最终确定它们由何种材料构成。与 GScan 的合作把该校在人工智能、机器学习和数据分析方面的专长,与该公司在开发先进扫描技术方面的经验结合在一起。
Roy 解释说:“我们的目标是教会人工智能更准确地解读复杂的测量数据,从而无需实际打开结构,就能识别隐藏物体、确定其位置,并更好地了解结构内部的材料。”
宇宙粒子有助于揭示结构物内部的秘密
每一分钟,大约有 10,000 个缪子轰击地球表面每平方米的面积。这些天然、无害的粒子诞生于来自太空的宇宙射线撞击大气层之时,它们能够穿透数米厚的混凝土、钢铁和岩石。缪子成像技术正是利用了这种能力。探测器会追踪每一个缪子进入和离开物体的过程。大多数缪子径直穿过,但遇到致密物质的那部分会被推离原有路径。这些微小的偏转揭示了物体内部的情况——无需打开物体,也无需像 X 射线扫描仪那样使用人工辐射。
问题在于,缪子到达的速度很慢,而且测量噪声较大,因此生成的图像往往模糊不清。这正是塔尔图大学的研究人员与 GScan 借助 AI 要解决的问题:教会计算机从并不完美的数据中绘出清晰的图像。
“处理缪子数据需要复杂的机器学习和数据科学解决方案,”GScan 产品开发工程师 Tarvo Metspalu 说。“大学的研究帮助我们避免在产品开发中走入死胡同,并提升我们技术的准确度,让我们能够在不造成损坏的前提下评估建筑物和结构物的内部情况。”
机器学习提升物体定位的准确度
Roy 及其同事意识到,径直穿过物体的缪子与发生偏转的缪子同样具有信息价值。路径笔直,说明途中没有致密物质;路径偏转,则暗示存在致密物质。一个显示物体不在哪里,另一个显示物体可能在哪里。
研究人员构建了能同时读取这两类线索的机器学习模型。在计算机模拟中,这些模型定位隐藏物体的准确率高于仅使用单一类型数据的模型,也高于μ子成像中使用的标准方法。相关结果发表在《IEEE Access》期刊上。
研究人员还开发了区分彼此距离很近的物体的方法——在μ子图像中,这些物体可能看起来像一个单一的大物体。他们开发的混合方法在模拟中将错误合并物体的比例从约15%降至2%。"这一结果很重要,因为在实际应用中,我们需要能够区分物体位置和更小的细节,并检测结构内部的潜在变化,"Roy补充道。
应用范围从结构检测到材料识别
除了定位物体,塔尔图大学的研究人员正在研究从μ子测量中识别不同材料的方法。迄今为止,他们已分析了14种材料的特性,并测试了统计和机器学习方法来区分它们。如果 AI 不仅能识别物体的位置,还能识别其材质,这项技术或可用于检查工业设施,或探测危险和违禁材料。
下一个目标是提升模型的可靠性,并进一步开发用于训练它们的数据集,使 AI 能够处理更广泛的材料和更复杂的测量条件。尽管要将这项技术更广泛地用于评估建筑物和构筑物的状况,仍需进一步开发与测试,但塔尔图大学与 GScan 的合作表明:把物理测量与 AI 结合起来,能够为探测那些内部此前难以检查的结构开辟新的可能。
Kirill Rodionov 等,《基于机器学习的直线与散射缪子射线双视角目标定位》,IEEE Access(2026)。DOI: 10.1109/access.2025.3648859
期刊信息:IEEE Access
这篇报道由谁撰写?
英语硕士,自 2021 年起担任文字编辑,拥有高等教育与健康类内容经验。致力于可信赖的科学新闻。完整简介 →
数学生物学学士,创意写作硕士。游历广泛,对科学与语言有独到见解。完整简介 →
引用:AI 改进宇宙粒子扫描,在模拟中更精准地定位隐藏物体(2026 年 11 月 12 日),2026 年 11 月 12 日检索自 https://techxplore.com/news/2026-11-ai-cosmic-particle-scans-hidden.html
本文件受版权保护。除为私人学习或研究目的的合理使用外,未经书面许可,不得复制任何部分。内容仅供参考。
混凝土梁的CT分析
AI帮助显微镜找到样本中最具信息量的纳米尺度特征
无形的触觉:研究人员赋予AI感知和测量表面的能力
人工智能解读探测器“云”以加速材料研究
超越物理传感器:AI扩散模型可视化隐藏的结构缺陷
研究人员解开关于辐射的物理学难题,有望带来更耐用的电子产品
早期测试中,手势与振动腰带让无人机无需视觉反馈即可受控
研究将水泥颗粒形状与建筑成本联系起来
电场可改变半导体的散热方式
借助超级计算机,飞机仿真攀上新高度
无线电波为更优的无线电动汽车充电铺平道路
获取即时摘要文本(Gist)
在仿真中,结合直线与偏转μ子路径的机器学习,比单一数据模型和标准μ子成像更准确地定位了隐藏物体。一种混合方法将相邻物体被错误合并的比例从约15%降至2%。实际的结构检测仍需进一步开发和测试。
本摘要由 LLM 自动生成。完整免责声明
译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。
查看英文原文(AI improves cosmic particle scans to locate hidden objects more accurately in simulations)
edited by Gaby Clark, reviewed by Robert Egan
This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility:
peer-reviewed publication
Add to Preferred Sources
Multiple coulomb scattering of a muon trajectory, with the scattering angle. This figure demonstrates a potential application: inspecting the interiors of trucks to detect concealed or prohibited materials. Credit: IEEE Access (2026). DOI: 10.1109/access.2025.3648859
Researchers at the University of Tartu Institute of Computer Science are working with the Estonian technology company GScan to develop AI-based methods that use measurement data collected with the help of cosmic particles to locate objects more accurately. The technology is useful for assessing the condition of bridges and other large structures and identifying potential damage.
At the University of Tartu's Institute of Computer Science, lecturer Kallol Roy and his research colleagues are developing intelligent methods to analyze information collected by a special scanning technology known as muon tomography. The aim is to help computers determine what is hidden inside large objects, where those objects are located and, eventually, what materials they contain. The collaboration with GScan brings together the university's expertise in artificial intelligence, machine learning and data analysis with the company's experience in developing advanced scanning technologies.
"Our goal is to teach artificial intelligence to interpret complex measurements more accurately, so that we can identify hidden objects, determine their locations and better understand the materials inside structures without physically opening them," explained Roy.
Cosmic particles help reveal what lies inside structures
Every minute, about 10,000 muons strike each square meter of the Earth's surface. These natural, harmless particles are created when cosmic rays from space hit the atmosphere, and they can pass through meters of concrete, steel and rock. Muon tomography puts this ability to work. Detectors track each muon as it enters and leaves an object. Most fly straight through, but those that meet something dense are nudged off course. These tiny deflections reveal what lies inside—without opening the object and without the artificial radiation an X-ray scanner needs.
The catch is that muons arrive slowly and the measurements are noisy, so the resulting images are often blurry. This is the problem the University of Tartu researchers and GScan are tackling with AI: teaching computers to draw a clear picture from imperfect data.
"Processing muon data requires complex machine learning and data science solutions," says Tarvo Metspalu, product development engineer at GScan. "Research at the university helps us avoid dead ends in product development and improve the accuracy of our technology, so that we can assess the inside of buildings and structures without damaging them."
Machine learning improves the accuracy of object localization
Roy and his colleagues realized that the muons passing straight through an object are just as informative as those that are deflected. A straight path suggests that nothing dense stands in the way; a deflected one hints that something does. One shows where an object is not; the other shows where it is likely to be.
The researchers built machine learning models that read both kinds of clues at once. In computer simulations, these models located hidden objects more accurately than models using only one type of data, and more accurately than the standard method used in muon imaging. The results are published in the journal IEEE Access.
The researchers have also developed methods to distinguish between closely spaced objects, which may appear as a single large object in muon images. The hybrid method they developed reduced the proportion of incorrectly merged objects in simulations from approximately 15% to 2%. "This result is important because, in practical applications, we need to be able to distinguish object locations and smaller details and detect potential changes inside structures," Roy added.
Applications range from structural inspection to material identification
In addition to locating objects, researchers at the University of Tartu are investigating ways to identify different materials from muon measurements. So far, they have analyzed the properties of 14 materials and tested statistical and machine learning methods to distinguish between them. If AI can identify not only an object's location but also its material, the technology could be useful for inspecting industrial facilities or detecting hazardous and prohibited materials.
The next goal is to improve the reliability of the models and further develop the datasets used to train them, enabling AI to handle a wider range of materials and more complex measurement conditions. Although wider use of the technology for assessing the condition of buildings and structures still requires further development and testing, the collaboration between the University of Tartu and GScan demonstrates how combining physical measurements with AI can open up new possibilities for examining structures whose interiors have previously been difficult to inspect.
Kirill Rodionov et al., ML-Based Dual View Object Localization From Straight and Scattered Muon Rays, IEEE Access (2026). DOI: 10.1109/access.2025.3648859
Journal information:IEEE Access
Who's behind this story?
MA in English, copy editor since 2021 with experience in higher education and health content. Dedicated to trustworthy science news. Full profile →
Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile →
Citation: AI improves cosmic particle scans to locate hidden objects more accurately in simulations (2026, November 12) retrieved 12 November 2026 from https://techxplore.com/news/2026-11-ai-cosmic-particle-scans-hidden.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.
CT analysis of concrete beams
AI helps microscopes find the most informative nanoscale features in a sample
Invisible touch: Researchers give AI the ability to feel and measure surfaces
Artificial intelligence deciphers detector 'clouds' to accelerate materials research
Beyond physical sensors: AI diffusion models visualize hidden structural defects
Researchers solve a physics puzzle about radiation that could lead to tougher electronics
Hand gestures and vibrating belt enable drone control without visual feedback in early tests
Study links cement particle shape and construction costs
Electric fields can change how semiconductors dissipate heat
Aircraft simulations scale new heights with supercomputer
Radio waves pave the way for better wireless EV charging
Get Instant Summarized Text (Gist)
In simulations, machine learning combining straight and deflected muon paths localized hidden objects more accurately than single-data models and standard muon imaging. A hybrid method reduced incorrect merging of nearby objects from about 15% to 2%. Practical structural inspection requires further development and testing.
This summary was automatically generated using LLM. Full disclaimer
This site uses cookies to assist with navigation, analyse your use of our services, collect data for ads personalisation and provide content from third parties. By using our site, you acknowledge that you have read and understand our Privacy Policy and Terms of Use.
原文为英文,正文由自动抓取并保留原文。
原文发布于 TechXplore(AI improves cosmic particle scans to locate hidden objects more accurately in simulations),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。
