基于三维荧光光谱的原酒品质评价模型建立

doi:10.3964/j.issn.1000-0593(2024)12-3391-08

摘要
参考文献
相关文章 (15)

全文: PDF (5948 KB)
输出: BibTeX | EndNote (RIS)

摘要：酒厂的生产主要采用传统的“看花摘酒”工艺，依靠工人的主观经验进行评价。实际生产中受到诸多因素的影响，导致接酒过程的不确定性，原酒品质的稳定性难以得到保证。通过采集原酒样本并配制掺加不同浓度酒尾（0.0%~2.0%）的复合原酒，进行荧光扫描得到三维荧光光谱，建立物质变化与荧光数据变化之间的联系：采取切除散射、拉曼归一化、Savitzky-Golay平滑、去除异常值等方法进行光谱预处理，通过平行因子分析将其分解为四个互不相关的组分，结合单物质荧光光谱特点进行综合相似度分析，对各组分进行了初步鉴别。结果表明，大部分酸类和酯类物质的荧光光谱与组分二相关性更大，组分二的荧光特性受酸类和酯类物质影响更大；数据集大小由781×61×164简化为4×164，达到了数据降维的效果。建立了支持向量机模型（SVM）对原酒品质进行评价，采取遗传寻优算法（GA）对支持向量机模型优化。GA-SVM模型较原始SVM模型性能有所提升，优化后的模型准确率由88.64%提升至95.45%，模型精确率由0.94提升至1.00。三维荧光结合化学计量学可以作为一种快速检测的有效手段对原酒质量进行评价，为酒厂摘酒过程实现在线检测提供支持。

关键词：白酒；三维荧光；相似性；平行因子分析；支持向量机

Abstract：At present, traditional liquor selection commonly employs the method of “liquor picking by flowers” during production, relying on workers' subjective experience for evaluation. However, multiple influencing factors affect the actual production, resulting in uncertainty in the process of liquor connection, posing challenges in ensuring the stability of original liquor quality. This study collected samples of the original and composite original liquor with varying concentrations of tail liquor (0.0%~2.0%). The three-dimensional fluorescence spectra were obtained by fluorescence scanning, establishing a correlation between substance changes and fluorescence data. The fluorescence spectra underwent pre-processing steps such as removing scattering, Raman normalization, Savitzky-Golay smoothing, and removing outliers. Subsequently, parallel factor analysis was used to decompose the spectra into four uncorrelated components, and these components were initially identified through composite similarity analysis in conjunction with the attributes observed in single-substance fluorescence spectra. The results show a higher correlation between the fluorescence spectra of most acids and esters with component 2, suggesting that acids and esters have a stronger influence on the fluorescence properties of component 2. The dataset is reduced from 781×61×164 to 4×164, achieving data dimensionality reduction. A support vector machine (SVM) model was developed to assess the quality of the original liquor. A genetic algorithm (GA) was also employed to optimize the SVM model. GA-SVM model performs better than the original SVM model in accuracy and precision. The optimized model achieved an accuracy of 88.64% compared to 95.45% of the original model, and the precision improved from 0.94 to 1.00. This suggests that integrating 3-D fluorescence and chemometrics is an effective method for rapid detection to evaluate the quality of the original liquor. And provide support for online detection during the distillate liquor selection process, thereby enhancing the overall quality control.

Key words：Baijiu; Three-dimensional fluorescence; Similarity; Parallel factor analysis; Support vector machine

收稿日期: 2023-11-15 修订日期: 2024-02-21

中图分类号:

O657.3

基金资助: 国家重点研发计划项目（2022YFC3401300），国家自然科学基金项目（32371540，21878128）资助

通讯作者: 杨海麟 E-mail: yanghailin@jiangnan.edu.cn

作者简介: 孙雍荣，女，1996年生，江南大学生物工程学院与工业生物技术教育部重点实验室硕士研究生
e-mail: 6210208050@stu.jiangnan.edu.cn

引用本文:

孙雍荣，权志熙，丁林志，冯守帅，龙凌凤，杨海麟. 基于三维荧光光谱的原酒品质评价模型建立[J]. 光谱学与光谱分析, 2024, 44(12): 3391-3398.
SUN Yong-rong, QUAN Zhi-xi, DING Lin-zhi, FENG Shou-shuai, LONG Ling-feng, YANG Hai-lin. A Model for Evaluating the Quality of Original Liquor Using Three-Dimensional Fluorescence Spectra. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2024, 44(12): 3391-3398.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2024)12-3391-08 或 https://www.gpxygpfx.com/CN/Y2024/V44/I12/3391

[1] SHEN Yi-fang（沈怡芳）. Liquor Technical Manual（白酒生产技术全书）. Beijing: China Light Industry Press（北京：中国轻工业出版社）, 2007.
[2] ZHANG Kun, MA Rong-shan, CHENG Shi, et al（张堃, 马荣山, 程什, 等）. China Brewing（中国酿造）, 2010, (9): 36.
[3] Sádecká J, Uríčková V, Jakubíková M. Applications of Molecular Spectroscopy to Current Research in the Chemical and Biological Sciences, 2016, 339: 362.
[4] Cozzolino Daniel, Cynkar Wies, Shah Nevil, et al. Food Analytical Methods, 2011, 5(3): 381.
[5] Zhou X J, Li L, Zheng J, et al. Food Chemistry, 2024, 436: 137739.
[6] Kellner R, Mermet J M, Otto M, et al. Analytical Chemistry, 2006, 385(6): 1033.
[7] Geng T, Wang Y, Yin X L, et al. Critical Reviews in Analytical Chemistry, 2023, 1: 23.
[8] Qian C, Chen W, Li W H, et al. Frontiers of Environmental Science & Engineering, 2017, 11(4): 8.
[9] Lenhardt Lea, Bro Rasmus, Zekovic Ivana, et al. Food Chemistry, 2015, 175: 284.
[10] Peng T Q, Yin X L, Sun W, et al. Food Analytical Methods, 2019, 12: 2306.
[11] Hu X C, Yu H, Deng Y, et al. European Food Research and Technology, 2023, 249(3): 767.
[12] Song J Y, Gu H W, Wang Y, et al. European Food Research and Technology, 2023, 249(8): 2087.
[13] Wu Q, Geng T, Yan M L, et al. Journal of Food Composition and Analysis, 2024, 125: 105763.
[14] Wu J H, Peng H B, Li L, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2023, 284: 121790.
[15] Lawaetz A J, Stedmon C A. Applied Spectroscopy, 2009, 63(8): 936.
[16] Dramicanin T, Zekovic I, Perisa J, et al. Journal of Fluorescence, 2019, 29(5): 1103.
[17] Liu B B, Wu H L, C Y, et al. Analytical Methods, 2023, 15(4): 502.
[18] Kruse F A, Lefkoff A B, Boardman J W, et al. Remote Sensing of Environment, 1993, 44(2-3): 145.
[19] Facci Rômulo R, Cezário Paulo S. De O, et al. Analytical Letters, 2020, 54(9): 1522.
[20] JIN Xi-qing, YIN Hong, ZHAN Song-qiang, et al（靳喜庆，殷红，谌松强，等）. China Brewing（中国酿造）, 2023, 42(1): 203.
[21] FAN Wen-lai, NIE Qing-qing, XU Yan（范文来, 聂庆庆, 徐岩）. Food Science（食品科学）, 2013, 34(4): 135.
[22] He F, Yang S Q, Zhang G H, et al. Journal of Food Composition and Analysis, 2022, 110: 104577.
[23] Yu S H, Xiao X, Wang Z G, et al. Applied Spectroscopy, 2011, 65(4): 369.