基于近红外光谱结合机器算法的新陈香榧籽无损鉴别

doi:10.19894/j.issn.1000-0518.240429

摘要/Abstract

摘要：

快速无损鉴别香榧陈籽和新籽，对保护消费者权益十分重要。采用可见光-近红外光谱仪（光谱仪Ⅰ）、近红外中短波光谱仪（光谱仪Ⅱ）采集带壳香榧样本的光谱，使用9种方法对光谱数据进行预处理，然后利用区间优化选择算法（ICO）、竞争自适应重加权采样（CARS）、连续投影算法（SPA）和变量组合种群分析（VCPA）选择方法筛选显著相关的光谱特征变量，基于优化后的特征变量，分别应用线性判别分析（LDA）、支持向量机（SVM）和反向传播神经网络（BP）方法建立香榧陈籽的判别模型。结果表明，通过光谱特征优化与模型适配，对于可见-近红外光谱，CARS-SVM模型性能较优，其预测集的敏感性、特异性和准确率均为100%；对于近红外中短波光谱，标准正态变量变换（SNV）预处理联合VCPA优化特征变量构建的BP模型较优，其预测集敏感性、特异性和准确率分别为98.18%、93.02%和95.04%。由此可知，近红外光谱结合化学计量学方法能较好地实现香榧陈籽的判别。旨为香榧陈籽快速无损判别提供一种新的检测方法，有效区分香榧的品质。

关键词: 近红外光谱, 香榧籽, 香榧陈籽, 波长变量选择, 判别模型

Abstract:

Rapid and non-destructive identification of staled and fresh Torreya grandis seeds is crucial for protecting consumers' rights and interests two NIR spectral instruments，visible-near infrared （spectrometer Ⅰ） and short-medium wave NIR （spectrometer Ⅱ）， were employed to acquire spectral data from shelled Torreya grandis samples. Nine preprocessing methods were applied to the spectral data， followed by feature variable selection using four algorithms： interval combination optimization （ICO）， competitive adaptive reweighted sampling （CARS）， successive projections algorithm （SPA）， and variable combination population analysis （VCPA）. Based on optimized variables， three classification models-linear discriminant analysis （LDA）， support vector machine （SVM）， and backpropagation neural network （BP） were established. Results demonstrated that spectral feature optimization significantly enhanced model performance. For visible-NIR spectra， the CARS-SVM model achieved optimal classification， with sensitivity， specificity， and accuracy rates of 100% in the prediction set. For short-medium wave NIR spectra， the BP neural network model incorporating standard normal variate （SNV） preprocessing and VCPA-optimized variables exhibited superior performance， attaining sensitivity， specificity， and accuracy rates of 98.18%， 93.02% and 95.04%， respectively. It is aim to provide a detection method for rapid and non-destructive identification of Torreyagrandis stale seeds， effectively distinguishing their quality.

Key words: Near-infrared spectroscopy, Torreya grandis seeds, Torreya grandis stale seeds, Wavelength variable selection, Discrimination model

中图分类号:

O657

范郑欣, 昝佳君, 杜园, 孙通. 基于近红外光谱结合机器算法的新陈香榧籽无损鉴别[J]. 应用化学, 2025, 42(6): 838-849.

Zheng-Xin FAN, Jia-Jun ZAN, Yuan DU, Tong SUN. Rapid and Non-Destructive Identification of New Torreya grandis Seeds Based on Near-Infrared Spectroscopy and Feature Wavelength Selection Combined with Machine Algorithm for Discrimination[J]. Chinese Journal of Applied Chemistry, 2025, 42(6): 838-849.

图/表 12

参考文献 31

1	王蕤，王玫鹃，汤富彬，等. 浙江省不同地区香榧生籽营养品质差异性分析与评价［J］. 核农学报， 2021， 35（11）： 2578-2588.
	WANG R， WANG M J， TANG F B， et al. Analysis and evaluation of nutritional quality differences of Torreya grandis in different regions of Zhejiang Province［J］. J Nuclear Agric Sci， 2021， 35（11）： 2578-2588.
2	何慈颖，娄和强，吴家胜. 香榧油脂及其合成调控机制研究进展［J］. 浙江农林大学学报， 2023， 40（4）： 714-722.
	HE C Y， LOU H Q， WU J S. Research progress on the oil and its synthesis regulation mechanism of Torreya grandis［J］. J Zhejiang A&F Univ， 2023， 40（4）： 714-722.
3	LI H L， XIAO N， SUN T， et al. Influence of lighting pattern and sample positioning on detection of moldy core disease in apples by NIR spectroscopy［J］. Food Bioprocess Technol， 2024， 17（12）： 5221-5241.
4	CORTÉS V， BLASCO J， ALEIXOS N， et al. Monitoring strategies for quality control of agricultural products using visible and near-infrared spectroscopy： a review［J］. Trends Food Sci Technol， 2019， 85： 138-148.
5	QIAO L， MU Y C， LU B， et al. Calibration maintenance application of near-infrared spectrometric model in food analysis［J］. Food Rev Int， 2023， 39（3）： 1628-1644.
6	SQUEO G， CRUZ J， DE ANGELIS D， et al. Considerations about the gap between research in near-infrared spectroscopy and official methods and recommendations of analysis in foods［J］. Curr Opin Food Sci， 2024， 59： 101203.
7	VELEZ-SILVA N L， DRENNEN J K， ANDERSON C A. Continuous manufacturing of pharmaceutical products： a density-insensitive near infrared method for the in-line monitoring of continuous powder streams［J］. Int J Pharm， 2024， 650： 123699.
8	REN S H， JIA Y F. Near-infrared data classification at phone terminal based on the combination of PCA and CS-RBFSVC algorithms［J］. Spectrochim Acta A， 2023， 287： 122080.
9	LOU Q J， LEI M， WANG Y， et al. Diagnostic features emerging in near-infrared reflectance spectroscopy for low petroleum hydrocarbon pollution after spectral subtraction［J］. SCI Total Environ， 2024， 928： 172264.
10	LIU S Y， WANG S T， HU C H， et al. apid and accurate determination of diesel multiple properties through NIR data analysis assisted by machine learning［J］. Spectrochim Acta A， 2022， 277： 121261.
11	HU J Q， MA X C， LIU L L， et al. Rapid evaluation of the quality of chestnuts using near-infrared reflectance spectroscopy［J］. Food Chem， 2017， 231： 141-147.
12	刘洁. 基于近红外光谱技术的板栗品质无损检测方法研究［D］. 武汉：华中农业大学， 2011.
	LIU J. Research on nondestructive detection method of chestnut quality based on near infrared spectroscopy technology［D］. Wuhan： Huazhong Agricultural University， 2011.
13	张严，谢岩黎，孙淑敏，等. 近红外分析花生籽粒脂肪酸含量的研究［J］. 河南工业大学学报（自然科学版）， 2014， 35（2）： 54-58.
	ZHANG Y， XIE Y L， SUN S M， et al. Research on near infrared analysis of fatty acid content in Peanut Grains［J］. J Henan Univ Technol （Nat Sci Ed）， 2014， 35（2）： 54-58.
14	CANNEDDU G， JÚNIOR L C， DE ALMEIDA TEIXEIRA G H. Quality evaluation of shelled and unshelled macadamia nuts by means of near-infrared spectroscopy （NIR）［J］. J Food Sci， 2016， 81（7）： C1613-C1621.
15	MOSCETTI R， MONARCA D， CECCHINI M， et al. Detection of mold-damaged chestnuts by near-infrared spectroscopy［J］. Postharvest Bio Techol， 2014， 93： 83-90.
16	CARVALHO L C， MORAIS C L M， LIMA K M G， et al. Using intact nuts and near infrared spectroscopy to classify macadamia cultivars［J］. Food Anal Method， 2018， 11（7）： 1857-1866.
17	GHOSH S， MISHRA P， MOHAMAD S N H， et al. Discrimination of peanuts from bulk cereals and nuts by near infrared reflectance spectroscopy［J］. Biosyst Eng， 2016， 151： 178-186.
18	HUANG T， CAI G， LIU H， et al. Selecting near-infrared reflection spectroscopy pretreatment methods by chemical components valid and invalid absorption wavebands［J］. Spectr Lett， 2022， 55（9）： 607-617.
19	KONG L F， WU C Z， LI H L， et al. Discrimination of tea seed oil adulteration based on near-infrared spectroscopy and combined preprocessing method［J］. J Food Compos Anal， 2024， 134： 106560.
20	ZHANG W， PAN L， LU L. Prediction of TVB-N content in beef with packaging films using visible-near infrared hyperspectral imaging［J］. Food Control， 2023， 147： 109562.
21	LI M X， SHI Y B， ZHANG J B， et al. Rapid evaluation of Ziziphi Spinosae Semen and its adulterants based on the combination of FT-NIR and multivariate algorithms［J］. Food Chem： X， 2023， 20： 101022.
22	TANG R， CHEN X， LI C. Detection of nitrogen content in rubber leaves using near-infrared （NIR） spectroscopy with correlation-based successive projections algorithm （SPA）［J］. Appl Spectrosc， 2018， 72（5）： 740-749.
23	JIANG H， XU W， CHEN Q. Comparison of algorithms for wavelength variables selection from near-infrared （NIR） spectra for quantitative monitoring of yeast （Saccharomyces cerevisiae） cultivations［J］. Spectrochim Acta A， 2019， 214： 366-371.
24	LI M， XIA L， WU Q， et al. MD-LDA： a supervised LDA topic model for identifying mechanism of disease in TCM［J］. Data Technol Appl， 2024， 59（1）： 1-18.
25	ALMOUJAHED M B， RANGARAJAN A K， WHETTON R L， et al. Non-destructive detection of fusarium head blight in wheat kernels and flour using visible near-infrared and mid-infrared spectroscopy［J］. Chemom Intell Lab， 2024， 245： 105050.
26	SCHREUDER J， NIKNAFS S， WILLIAMS P， et al. Non-destructive prediction of fertility and sex in chicken eggs using the short wave near-infrared region［J］. Spectrochim Acta A， 2024， 322： 124716.
27	ZHANG X， SUN J， LI P， et al. Hyperspectral detection of salted sea cucumber adulteration using different spectral preprocessing techniques and SVM method［J］. LWT， 2021， 152： 112295.
28	MAO S， ZHOU J， HAO M， et al. BP neural network to predict shelf life of channel catfish fillets based on near infrared transmittance （NIT） spectroscopy［J］. Food Packaging Shelf， 2023， 35： 101025.
29	TAN B， YOU W， HUANG C， et al. An intelligent near-infrared diffuse reflectance spectroscopy scheme for the non-destructive testing of the sugar content in cherry tomato fruit［J］. Electronics， 2022， 11（21）： 3504.
30	刘婷婷. 基于近红外光谱的猕猴桃糖度分级及品种鉴别技术研究［D］. 西安：西京学院， 2024.
	LIU T T. Research on Kiwifruit agar classification and varitty classification technology based on near infrared spectroscopy［D］. Xi'an： Xijing University， 2024.
31	余展旺，殷海，何曼文，等. 基于高光谱成像技术对菜心种子霉变的识别［J］. 种子， 2024， 43（8）： 146-150， 156.
	YU Z W， YIN H， HE M W， et al. Identification of cabbage seed mildew based on hyperspectral imaging technology［J］. Seeds， 2024， 43（8）： 146-150， 156.

编辑推荐 0

Metrics

阅读次数

全文

HTML			PDF

最新录用	在线预览	正式出版	最新录用	在线预览	正式出版
0	0	2	0	0	34

来源	本网站	其他网站

次数	36	0
比例	100%	0%

摘要

最新录用	在线预览	正式出版

0	0	99

	来源	本网站

	次数	99
	比例	100%

Principal component	Spectrometer Ⅰ		Spectrometer Ⅱ
Principal component	Variance contribution rate/%	Cumulative contribution rate/%	Variance contribution rate/%	Cumulative contribution rate/%
1	84.44	84.44	79.21	79.21
2	14.44	98.89	15.12	94.33
3	0.84	99.73	3.92	98.25
4	0.21	99.94	1.09	99.34

Principal component	Spectrometer Ⅰ		Spectrometer Ⅱ
Principal component	Variance contribution rate/%	Cumulative contribution rate/%	Variance contribution rate/%	Cumulative contribution rate/%
1	84.44	84.44	79.21	79.21
2	14.44	98.89	15.12	94.33
3	0.84	99.73	3.92	98.25
4	0.21	99.94	1.09	99.34

Data set	Spectral preprocessing	Modeling methods	Calibration set			Prediction set
Data set	Spectral preprocessing	Modeling methods	Sensitivity/%	Specificity/%	Accuracy/%	Sensitivity/%	Specificity/%	Accuracy/%
Spectrometer Ⅰ	None	LDA	99.34	97.67	98.58	73.13	98.67	86.62
	None	SVM	99.22	98.69	98.93	100.00	98.25	99.29
	None	BP	98.44	98.04	98.22	100.00	98.25	99.29
Spectrometer Ⅱ	Standard	LDA	92.31	91.89	92.09	90.12	96.67	92.91
	SNV	SVM	89.12	87.02	88.13	89.29	88.24	88.65
	SNV	BP	88.36	85.61	87.05	90.91	88.37	89.36

Data set	Spectral preprocessing	Modeling methods	Calibration set			Prediction set
Data set	Spectral preprocessing	Modeling methods	Sensitivity/%	Specificity/%	Accuracy/%	Sensitivity/%	Specificity/%	Accuracy/%
Spectrometer Ⅰ	None	LDA	99.34	97.67	98.58	73.13	98.67	86.62
	None	SVM	99.22	98.69	98.93	100.00	98.25	99.29
	None	BP	98.44	98.04	98.22	100.00	98.25	99.29
Spectrometer Ⅱ	Standard	LDA	92.31	91.89	92.09	90.12	96.67	92.91
	SNV	SVM	89.12	87.02	88.13	89.29	88.24	88.65
	SNV	BP	88.36	85.61	87.05	90.91	88.37	89.36

Wavelength selection methods	Modeling methods	Number of variables	Preprocessing	Calibration set			Prediction set
Wavelength selection methods	Modeling methods	Number of variables	Preprocessing	Sensitivity/%	Specificity/%	Accuracy/%	Sensitivity/%	Specificity/%	Accuracy/%
CARS	LDA	30	None	100.00	99.22	99.64	75.68	100.00	87.23
	SVM	30	None	100.00	100.00	100.00	100.00	100.00	100.00
	BP	30	None	100.00	100.00	100.00	100.00	98.25	99.29
ICO	LDA	172	None	100.00	100.00	100.00	75.68	98.51	86.52
	SVM	172	None	100.00	99.35	99.64	100.00	98.25	99.29
	BP	172	None	100.00	99.35	99.64	100.00	98.25	99.29
SPA	LDA	14	None	99.34	99.22	99.29	75.68	98.51	86.52
	SVM	14	None	99.22	99.34	99.29	100.00	98.25	99.29
	BP	14	None	96.85	96.10	96.44	97.65	96.43	97.16
VCPA	LDA	14	None	100.00	99.22	99.64	75.68	98.51	86.52
	SVM	14	None	100.00	99.35	99.64	100.00	98.25	99.29
	BP	14	None	100.00	100.00	100.00	98.82	98.21	98.58