基于X射线吸收光谱法的塑料薄膜厚度测量

doi:10.3964/j.issn.1000-0593(2023)11-3461-08

摘要
参考文献
相关文章 (15)

全文: PDF (3567 KB)
输出: BibTeX | EndNote (RIS)

摘要：塑料薄膜是我国塑料产品中占比五分之一的大宗类型，在厂家生产时最重要的指标之一为塑料薄膜厚度，如何准确、快速、方便地测量塑料薄膜厚度是一项具有重大经济价值的研究课题。为验证X射线吸收光谱法测量塑料薄膜厚度的可行性，制作了不同厚度的聚乙烯塑料薄膜实验样本，设置30 kV的管电压以及1 μA的管电流激发X射线，照射不同厚度的塑料薄膜样品，用X射线探测器采集空白光谱数据和不同样本的原始X射线吸收光谱数据，得到各光谱在256个通道中的光子强度。在数据分析过程中，为达到数据降维的效果，选择主成分分析法处理所采集的数据；再将维数降低后的新数据集分两次分析，一次直接进行机器学习，另一次进行归一化处理后再进行机器学习。在机器学习中，其中的70％作为训练集，剩余的30％作为测试集，输入数据为各组样本X射线吸收光谱，输出数据为模型预测的塑料薄膜厚度。同时，为降低随机性导致的误差，多次训练，以平均的准确率来评价厚度估计的效果。最后，对比分析实验数据的结论是，当误差包容度设置为50 μm时，使用归一化处理后经机器学习的X射线吸收光谱法测量塑料薄膜厚度的准确率可以达到98.4％。同时，只要增加原始光谱数据的样本数，并有效规划不同厚度的采样分布，理论上可以大大提高该方法的精度，而且可以推广到其他材料的测厚任务。与市场上的其他测厚方式相比，X射线吸收光谱法测厚具备无损检测、快速检测以及适用范围广的优势，这对于丰富厂家生产线以及相关监管部门的塑料薄膜测厚技术、提高测厚效率、提升测量准确率具有较好的应用前景。

关键词：X射线吸收光谱；塑料薄膜厚度；主成分分析；机器学习

Abstract：Plastic film is a bulk type accounting for one-fifth of plastic products in China. One of the most important indicators in manufacturers' production is the thickness of plastic film. How to accurately, quickly and conveniently measure the thickness of plastic film is a research topic with great economic value. In this paper, in order to verify the feasibility of measuring the thickness of the plastic film by X-ray absorption spectroscopy, experimental samples of polyethylene plastic film with different thicknesses are made, and the 30 kV pipe voltage and 1 μA. The tube current of A excites X-rays, irradiates plastic film samples of different thicknesses, collects blank spectral data and original X-ray absorption spectral data of different samples with X-ray detector, and obtains photon intensity of each spectrum in 256 channels. In the process of data analysis, in order to achieve the effect of data dimension reduction, principal component analysis is selected to process the collected data; The new dataset with reduced dimension is analyzed two times, one for machine learning directly and the other for machine learning after normalization. In machine learning, 70% are used as training sets, and the remaining 30% are used as test sets. The input data is the X-ray absorption spectra of each group of samples, and the output data is the plastic film thickness predicted by the model. At the same time, to reduce the error caused by randomness, multiple trainings were conducted to evaluate the effect of thickness estimation with average accuracy. Finally, comparative analysis of experimental data concludes that when the error tolerance is set to 50 μm, the accuracy of measuring the thickness of the plastic film by using the machine-learned X-ray absorption spectroscopy after normalization can reach 98.4%. At the same time, as long as the number of samples of the original spectral data is increased and the sampling distribution of different thicknesses is effectively planned, the accuracy of this method can be greatly improved in theory and can be extended to the thickness measurement task of other materials. Compared with other thickness measurement methods on the market, X-ray absorption spectroscopy has the advantages of nondestructive testing, rapid testing and a wide application range, which has a good application prospect for enriching the plastic film thickness measurement technology of manufacturers' production lines and relevant regulatory departments, improving the thickness measurement efficiency and improving the measurement accuracy. It has a good application prospect to enrich the thickness measuring technology of plastic film of the production line and related supervision department, improve the thickness measuring efficiency and accuracy.

Key words：X-ray absorption spectroscopy; Plastic film thickness measurement; Principal component analysis; Machine learning

收稿日期: 2022-10-07 修订日期: 2023-05-15

中图分类号:

O67.39

基金资助: 国家自然科学基金项目（62275223）资助

作者简介: 方正，1976年生，厦门大学仪器与电气系教授 e-mail: fangzheng@xmu.edu.cn

引用本文:

方正，王涵博. 基于X射线吸收光谱法的塑料薄膜厚度测量[J]. 光谱学与光谱分析, 2023, 43(11): 3461-3468.
FANG Zheng, WANG Han-bo. Measurement of Plastic Film Thickness Based on X-Ray Absorption Spectrometry. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2023, 43(11): 3461-3468.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2023)11-3461-08 或 https://www.gpxygpfx.com/CN/Y2023/V43/I11/3461

[1] XU Xing-xiang（徐兴祥）. Shanghai Packaging（上海包装），2010，4(4): 13.
[2] MA Zhao-rong，LIU You-sheng，ZHANG Qian-qian，et al（马兆嵘，刘有胜，张芊芊，等）. Asian Journal of Ecotoxicology（生态毒理学报），2020，15(4): 21.
[3] JIANG Feng-hua，XIE Lin-qing，SUN Cheng-jun，et al（蒋凤华，谢林青，孙承军，等）. Chinese Science Bulletin（科学通报），2021，66(13): 1571.
[4] CUI Wei-xin，WANG Ke-jian（崔卫鑫，王克俭）. Plastics Packaging（塑料包装），2020，30(3): 44.
[5] GUO Bin，DING Qiu，ZHANG Fang，et al（郭斌，丁虬，张芳，等）. Standard and Quality of Light Industry（轻工标准与质量），2020，5(5): 76.
[6] YUAN Wei-jie（袁维杰）. Process Automation Instrumentation（自动化仪表），2019，40(4): 52.
[7] LIU Yi-si（刘以思）. Nuclear Electronics and Detection Technology（核电子学与探测技术），1995，15(3): 161.
[8] LI Wen-tao，TIAN Jun（李文涛，田俊）. Nuclear Electronics and Detection Technology（核电子学与探测技术），2019，39(4): 468.
[9] XU Yuan，HUANG Jin-qiang，CUI Can，et al（徐源，黄金强，崔灿，等）. Physics Experimentation（物理实验），2012，32(5): 31.
[10] He P，Yu J K，Hong L H. Applied Mechanics and Materials，2017，865: 619.
[11] MO Chang-tao，HUAN Shuai，SU Hai-lin(莫长涛，郇帅，苏海林). Physical Experimentation(物理实验)，2016，36(1): 9.
[12] LI Jun，LAI Wan-chang，ZHOU Liang-ping（李军，赖万昌，周良平）. Guangdong Trace Element Science(广东微量元素科学)，2012，19(8): 62.
[13] CHEN Yong-de（陈涌得）. Light Industry Science and Technology（轻工科技），2018，34(4): 100.
[14] LIU Zhong-you，YANG Ben-li（刘忠友，杨本立）. Corrosion & Protection in Petrochemical Industry（石油化工腐蚀与防护），2011，28(5): 48.
[15] MA Kai-li，TENG Yang-yang，ZHANG Zheng（马凯丽，滕阳阳，张铮）. Heat Treatment of Metals（金属热处理），2019，44(S1): 680.
[16] PAN Xiang，ZHANG Yong-kang，CHANG Guo-qiang（潘翔，张永康，常国强）. Metrology & Measurement Technique（计量与测试技术），2020，47(10): 89.
[17] LU Yi-bo（陆一波）. China Metrology（中国计量），2012，46(2): 104.
[18] MA Ping，QIU Li-qin，ZHAO Yu-min，et al（马平，邱丽琴，赵玉民，等）. Metrology Technique（计量技术），2007，165(12): 6.
[19] HUANG Yu-sheng（黄余盛）. Metallurgical Collections（冶金丛刊），2005，4(4): 35.
[20] Yano J，Yachandra V K. J. Photosynth. Res.，2009，102: 241.
[21] MA Li-dun(马礼敦). Shanghai Measurement and Testing（上海计量测试），2007，34(6): 2.
[22] Hu B，Zhang X L，Quyang Q N，et al. Measurement，2016，93: 252.
[23] XIONG Zhi-bin（熊志斌）. Journal of Quantitative & Technological Economics（数量经济技术经济研究），2013，30(10): 138.
[24] XIE Jian-hong，SHI Jian，XI Xiao-wen，et al（谢建红，石剑，习小文，等）. Journal of Ordnance Equipment Engineering（兵器装备工程学报），2021，42(4): 228.
[25] Tanabe K，Tamura T，Uesaka H. Appl. Spectrosc.，1992，46(5): 807.
[26] QI Fang-zhong，LIN Shao-qian，YU Ting-ting，et al（綦方中，林少倩，俞婷婷，等）. Computer Applications and Software（计算机应用与软件），2020，37(1): 116.
[27] LI He，ZHAO Wen-jing，LUO Xue-song，et al（李贺，赵文静，罗雪松，等）. Journal of Dalian University of Technology（大连理工大学学报），2020，60(6): 635.
[28] YANG Yan-rong，DAI Yun（杨彦荣，戴云）. Acta Optica Sinica（光学学报），2017，37(3): 0333001.
[29] LIU Jian-wei，ZHAO Hui-dan，LUO Xiong-lin，et al（刘建伟，赵会丹，罗雄麟，等）. Acta Automatica Sinica（自动化学报），2020，46(6): 1090.