基于近红外光谱和多变量数据处理的鸡蛋蛋黄颜色无损判别研究

doi:10.3964/j.issn.1000-0593(2025)04-1015-07

摘要
参考文献
相关文章 (15)

全文: PDF (5136 KB)
输出: BibTeX | EndNote (RIS)

摘要：蛋黄颜色是鸡蛋品质的重要指标，消费者更喜欢购买蛋黄颜色较深的鸡蛋。通常将鸡蛋打开，通过罗氏比色扇对蛋黄颜色进行判别，无损判别蛋黄颜色的研究具有重要意义。针对不同颜色蛋壳的鸡蛋，进行蛋黄颜色无损判别方法研究，通过近红外光谱数据采集，采用化学计量法建立定性分类预测模型，对影响蛋黄颜色成分进行分析，找到谱图吸收峰对应官能团。采集了90个粉壳蛋和89个白壳蛋的近红外光谱数据，罗氏比色扇记录的蛋黄颜色用于建立定性分类模型目标颜色，将样本按2∶1分为校正集和预测集，分别对单种颜色蛋壳样本和混合颜色蛋壳样本建立了预测模型。采用线性（偏最小二乘法判别PLS-DA、线性判别分析LDA）和非线性(卷积神经网络CNN、极限学习机ELM)的方法建立了分类模型，运用多种预处理方法，采用CARS特征波长筛选方法对光谱数据筛选了176个波长点。不同颜色蛋壳混合样本采用CARS波长筛选法、MSC和二阶导数的预处理方法，建立的偏最小二乘法分类模型准确率最高达91.67%，LDA达到98.11%。对粉壳蛋单独进行建模时，建立的偏最小二乘分类模型测试集准确率达到100%。对白壳蛋单独进行建模时，建立的偏最小二乘分类模型准确率达到了96.67%，而LDA模型准确率则达到了100%。结果表明，线性分析方法更加能表征鸡蛋光谱数据蛋黄颜色的特征，适合蛋黄颜色无损检测。该方法不仅能满足消费者的需求，而且蛋黄颜色判别结果对养殖场饲料喂养及调控起指导作用。

关键词：近红外光谱；蛋黄颜色；偏最小二乘法；线性判别分析；特征波长筛选；数据预处理

Abstract：Yolk color is an important indicator of egg quality, and consumers prefer to buy eggs with darker yolk color. Currently, the commonly used method involves physically opening the egg to distinguish the yolk color using the Roche fan method, so the research on non-destructive identification of yolk color is significant. This paper mainly studies the non-destructive identification method of yolk color for eggs with different eggshell colors. The data is collected by near-infrared spectroscopy. Then, the qualitative classification prediction model is established by using a chemometry algorithm. The components affecting egg yolk color are analyzed to find the functional groups corresponding to the spectral absorption peak. This study collected the NIR spectral data of 90 pink and 89 white eggs using the Roche fan method to record yolk color and establish qualitative classification models. The samples were divided into correction sets and prediction sets according to 2∶1, and prediction models were established for single-color and mixed-color samples, respectively. Linear (partial least square discriminant PLS-DA, linear discriminant analysis LDA) and nonlinear (convolutional neural network CNN, extreme learning machine ELM) methods were used to establish the classification models along sidevarious pretreatment and wavelength screening methods. CARS feature wavelength screening method was used to screen 176 wavelength points of spectral data. Combining CARS wavelength screening, MSC, and second derivative pretreatment methods for 2 kinds of color eggshell samples, the accuracy of the test set reached 91.67% by the PLS-DA model. In contrast, the LDA model reached 98.11%. For the pink shell eggs, the accuracy of the test set is 100% by the PLS-DA model. For the white shell eggs, the accuracy of the PLS-DA model is 96.67%, while that of the LDA model is 100%. These results demonstrate the efficacy of linear methods in characterizing the egg yolk color from spectra. This method can not only meet the needs of consumers but also play a guiding role in feed feeding and control of farms.

Key words：Near Infrared Spectroscopy; Yolk color; Partial least square method; Linear discriminant analysis; Feature wavelength screening; Data reprocessing

收稿日期: 2024-05-09 修订日期: 2024-09-18

中图分类号:

O657.33

基金资助: 国家自然科学基金项目（41771357），天津市企业科技特派员项目（21YDTPJC00580），天津市科技重大专项与工程项目（18Z XBFNC00310）资助

通讯作者: 于亚萍 E-mail: yaping261@163.com

作者简介: 温裕宽，1998年生，天津农学院工程技术学院硕士研究生 e-mail: kuan717253@163.com

引用本文:

温裕宽，董桂梅，李留安，于晓雪，于亚萍. 基于近红外光谱和多变量数据处理的鸡蛋蛋黄颜色无损判别研究[J]. 光谱学与光谱分析, 2025, 45(04): 1015-1021.
WEN Yu-kuan, DONG Gui-mei, LI Liu-an, YU Xiao-xue, YU Ya-ping. Nondestructive Identification of Egg Yolk Color Based on Near Infrared Spectrum and Multivariate Data Processing. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(04): 1015-1021.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)04-1015-07 或 https://www.gpxygpfx.com/CN/Y2025/V45/I04/1015

[1] ZHENG Yan, XU Qing-cui, FAN Li-xia, et al（郑颜, 胥清翠, 范丽霞, 等）. Food and Nutrition in China(中国食物与营养), 2022, 28(1): 45.
[2] Rakonjac S, Bogosavljevic B S, Škrbic Z, et al. Biotechnology in Animal Husbandry, 2018, 34(3): 335.
[3] Pal M, Molnár J. International Journal of Food Science and Agriculture, 2021, 5(1): 180.
[4] Drabik K, Karwowska M, Wengerska K, et al. Animals, 2021, 11(5): 1204.
[5] Cendron F, Currò S, Rizzi C, et al. Animals, 2022, 13(1): 77.
[6] Antova G A, Gerzilov V T, Petkova Z Y, et al. Journal of the Science of Food and Agriculture, 2019, 99(13): 5890.
[7] Nuningtyas Y F, Chang H L, Lin M J, et al. Earth and Environmental Science, 2020, 478(1): 012050.
[8] Saleh A A, Gawish E, Mahmoud S F, et al. Sustainability, 2021, 13(8): 4503.
[9] Grčevic M, Kralik Z, Kralik G, et al. Journal of the Science of Food and Agriculture, 2019, 99(5): 2292.
[10] Milovanovic B, Tomovic V, Djekic I, et al. Journal of Food Measurement and Characterization, 2021, 15(6): 5097.
[11] Bovšková H, Míková K, Panovská Z. Czech Journal of Food Sciences, 2014, 32(3): 213.
[12] Joubrane K, Mnayer D, Hamieh T, et al. American Journal of Analytical Chemistry, 2019, 10(10): 488.
[13] Aboonajmi M, Saberi A, Abbasian Najafabadi T, et al. International Journal of Food Properties, 2016, 19(5): 1163.
[14] Chen H, Tan C, Lin Z. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2019, 206: 484.
[15] Abdel-Nour N, Ngadi M, Prasher S, et al. Food and Bioprocess Technology, 2011, 4: 731.
[16] Roger J M, Mallet A, Marini F. Molecules, 2022, 27(20): 6795.
[17] Mishra P, Biancolillo A, Roger J M, et al. Trends in Analytical Chemistry, 2020, 132: 116045.
[18] Lee L C, Liong C Y, Jemain A A. Chemometrics and Intelligent Laboratory Systems, 2017, 163: 64.
[19] Yun Y H, Li H D, Deng B C, et al. TrAC Trends in Analytical Chemistry, 2019, 113: 102.
[20] Fu J, Yu H D, Chen Z, et al. Infrared Physics & Technology, 2022, 125: 104231.
[21] Ng W, Minasny B, Malone B P, et al. Computers and Electronics in Agriculture, 2019, 158: 201.
[22] Chu X L, Yuan H F, Lu W Z. Progress in Chemistry, 2004, 16(4): 528.
[23] Brasil Y L, Cruz-tirado J P, Barbin D F. Food Control, 2022, 131: 108418.
[24] Cruz-tirado J P, Lucimar Da Silva Medeiros M, Barbin D F. Journal of Food Engineering, 2021, 306: 110643.