Qualitative Identification of Terahertz Time-domain Spectra Based on Analytic Hierarchy Process
-
摘要:
太赫兹时域光谱技术属于远红外光谱技术的一种,其光谱能够反映物质的内部特性,具有丰富的物理和化学信息,因此利用太赫兹波可对含氮元素的添加剂进行定性识别。层次分析法原用于解决评价类问题,本文将其引入太赫兹光谱定性分析领域,提出一种结合层次分析法的太赫兹时域光谱定性识别方法,并采集数据对其进行验证。文中以6种添加剂作为实验样品,首先对采集到的太赫兹时域光谱数据进行预处理,将其转化为由峰值、峰位、峰数和整体趋势组成的数据集;然后将数据划分为对比集和测试集,构建结合层次分析法的添加剂定性识别模型,并进行参数寻优。结果表明,基于单一因素(即整体趋势、峰值、峰位、峰数)的添加剂定性识别精度分别为80.23%、70.93%、67.44%、40.70%,而基于多因素的层次分析定性识别方法识别精度可提升至92.44%。此外,在数据预处理阶段对吸收谱数据进行二值化模糊表征,并作为整体趋势的数据集,将层次分析定性识别模型与此种数据表征法结合后,识别精度可提至94.19%。研究结果证明了结合层次分析法的定性识别算法的有效性,且该法步骤简单,无需训练,适用于小样本的快速定性检测。
Abstract:Terahertz time-domain spectroscopy (THz-TDS) is a type of far-infrared spectroscopy that reflects the internal characteristics of substances and provides rich physical and chemical information. Therefore, terahertz waves can be used to qualitatively identify food additives containing nitrogen. Hierarchical analysis, originally developed for solving evaluation-type problems, is introduced in this study to the field of qualitative analysis of terahertz spectra. This paper proposes and evaluates a qualitative identification method that combines THz-TDS with hierarchical analysis. In this study, six nitrogen-containing food additives were selected as experimental samples. First, the acquired terahertz time-domain spectral data were preprocessed and transformed into four datasets: peak values, peak positions, peak numbers, and overall spectral trends. Next, the data were divided into comparison and test sets to construct a qualitative identification model incorporating hierarchical analysis, followed by parameter optimization. The results indicated that the qualitative identification accuracy of additives based on single factors: overall trend, peak value, peak position, and peak number were 80.23%, 70.93%, 67.44%, and 40.70%, respectively. The multi-factor hierarchical analysis-based method improved the identification accuracy to 92.44%. In addition, by binarizing the fuzzy characterization of the absorption spectrum data during preprocessing and using it as the basis for assessing overall trends, the recognition accuracy increased to 94.19% when combined with the hierarchical analysis model. These results demonstrate the effectiveness of the proposed qualitative identification algorithm. The method is straightforward, does not require training, and is well-suited for rapid qualitative detection of small sample sets.
-
-
表 1 AHP相对重要性尺度
Table 1 AHP relative importance scale
Scale Define Clarification 1 Equally important Elements a and b are equally important for an attribute 3 Slightly important Element a is slightly more important than element b for an attribute 5 More important Element a is more important than element b for an attribute 7 Clearly important Element a is significantly more important than element b for an attribute 9 Special importance Element a is particularly more important than element b for an attribute 2, 4, 6, 8. Midpoint A compromise between the two neighbouring scales above The reciprocal of the above scales Inverse comparison For example, a compares to b as 3 and b compares to a as 1/3. 表 2 添加剂样片信息统计
Table 2 Statistical of additives samples
Sample name Category number Sample concentration /% Sample quantity L-Alanine S1 S1-1 S1-2 S1-3 10% 3 S1-4 S1-5 S1-6 25% 3 S1-7 S1-8 S1-9 35% 3 Benzoic acid S2 S2-1 S2-2 S2-3 10% 3 S2-4 S2-5 S2-6 25% 3 S2-7 S2-8 S2-9 35% 3 Melamine S3 S3-1 S3-2 S3-3 10% 3 S3-4 S3-5 S3-6 25% 3 S3-7 S3-8 S3-9 35% 3 2, 4-Hexadienoic acid S4 S4-1 S4-2 S4-3 10% 3 S4-4 S4-5 S4-6 25% 3 S4-7 S4-8 S4-9 35% 3 Sudan-Ⅰ S5 S5-1 S5-2 S5-3 10% 3 S5-4 S5-5 S5-6 25% 3 S5-7 S5-8 S5-9 35% 3 Xylitol S6 S6-1 S6-2 S6-3 10% 3 S6-4 S6-5 S6-6 25% 3 S6-7 S6-8 S6-9 35% 3 表 3 识别精度及其对应参数统计表(AHP方法)
Table 3 Identification accuracy and corresponding parameters statistical table
Data represen-tation Parameter weighting Identification accuracy/% Overall trend Peak value Peak
positionPeak number YS 1 0 0 0 80.23 YS 0.6 0.1 0.2 0.1 92.44 MH 1 0 0 0 93.60 MH 0.8 0.1 0.1 0.0 94.19 -
[1] 王芳, 张春红, 赵景峰, 等. 基于混合机器学习法的太赫兹波鉴别草种的研究[J]. 激光与光电子学进展, 2021, 58(3): 318-324. WANG F, ZHANG C H, ZHAO J F, et al. Identification of a grass species using a terahertz wave based on hybrid machine learning method[J]. Laser & Optoelectronics Progress, 2021, 58(3): 318-324.
[2] 胡晓华, 刘伟, 刘长虹, 等. 基于太赫兹光谱和支持向量机快速鉴别咖啡豆产地[J]. 农业工程学报, 2017, 33(9): 302-307. HU X H, LIU W, LIU C H, et al. Rapid identification of producing area of coffee bean based on terahertz spectroscopy and support vector machine[J]. Transactions of the Chinese Society of Agricultural Engineering, 2017, 33(9): 302-307
[3] 周博, 朱文魁, 王赵改, 等. 基于太赫兹时域光谱技术的烟草组分识别[J]. 农业工程学报, 2022, 38(10): 310-316. DOI: 10.11975/j.issn.1002-6819.2022.10.037 ZHOU B, ZHU W K, WANG Z G, et al. Identification of tobacco materials based on terahertz time-domain spectroscopy[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(10): 310-316. DOI: 10.11975/j.issn.1002-6819.2022.10.037
[4] 崔向伟, 沈韬, 刘英莉, 等. 小样本太赫兹光谱识别[J]. 激光与光电子学进展, 2021, 58(1): 321-328. CUI X W, SHEN T, LIU Y L, et al. Recognition of small-sample terahertz spectrum[J]. Laser & Optoelectronics Progress, 2021, 58(1): 321-328.
[5] 甘岚, 沈鸿飞, 王瑶, 等. 基于改进DCGAN的数据增强方法[J]. 计算机应用, 2021, 41(5): 1305-1313. GAN L, SHEN H F, WANG Y, et al. Data augmentation method based onimproved deep convolutional generative adversarial networks[J]. Journal of Computer Applications, 2021, 41(5): 1305-1313.
[6] 吴建. 以案例教学为核心的适应性学习系统的设计与实现[D]. 杭州: 浙江工业大学, 2009. WU Jian. The design and implementation of case-based learning in adaptive learning[D]. Hangzhou: Zhejiang University of Technology, 2009.
[7] 郑凤翥, 宁飞, 王惠林, 等. 光电系统伺服稳定平台可靠性定性分析[J]. 应用光学, 2022, 43(5): 853-858. ZHEN F Z, NING F, WANG H L, et al. Qualitative analysis of reliability on servo stabilizationplatform of electro-optical system[J]. Journal of Applied Optics, 2022, 43(5): 853-858.
[8] 燕芳, 刘同华, 张俊林. 糖类同分异构体的太赫兹吸收峰形成机理研究[J]. 光学学报, 2022, 42(5): 228-234. YAN F, LIU T H, ZHANG J L. Formation mechanism of terahertz absorption peaks of carbohydrate isomers[J]. Acta Optica Sinica, 2022, 42(5): 228-234.
[9] Dorney T D, Baraniuk R G, Daniel M, etal. Material parameter estimation with terahertz time-domain spectroscopy[J]. Journal of the Optical Society of America A, 2001, 18(7): 1562.
[10] Duvillare L, Garet F, Coutaz J L, etal. A reliable method for extraction of material parameters in terahertz time-domain spectroscopy[J]. IEEE Journal of Selected Topics in Quantum Electronics, 1996, 2(3): 39.