To achieve the real-time online monitoring of rapeseed oil oxidative stability, a custom-built portable near-infrared and Raman spectroscopy (NIR-Raman) device combined with spectral fusion techniques was utilized to assess the oxidative stability of rapeseed oil quantitatively. The NIR and Raman spectral data were collected from rapeseed oil samples exhibiting varying degrees of oxidation. Eight preprocessing methods and three feature selection approaches were applied to construct partial least squares regression (PLSR) models using data-level, feature-level decision and multi-block fusion strategies. The accuracy of the models was evaluated using the prediction set determination coefficient (Rp2) and the prediction root mean square error (RMSEP). The results demonstrated that: 1) Taking peroxide value (POV) as an example, in the NIR spectrum, the SNV+MSC preprocessing method displayed the best performance with an Rp2 of 0.84 and RMSEP of 1.05; In the Raman spectrum, the SNV preprocessing method was the most optimal with an Rp2 of 0.85 and an RMSEP of 0.99; 2) Among the spectral fusion schemes, the sequential orthogonal partial least squares (SO-PLS) method with multi-block fusion outperformed the other two methods, achieving R2 of 0.978 and RMSEP of 0.392; 3) The prediction of rapeseed oil oxidation indicators, including malondialdehyde (MDA), anisidine value (AV), and conjugated diene value (CDV), using SO-PLS integrated NIR and Raman spectroscopy, R2 values reached 0.98, 0.98, and 0.97, respectively, and RMSEP values were 0.871, 1.039, and 2.132, respectively. In summary, this study applied the SO-PLS fusion of NIR and Raman spectra for rapid, non-destructive, real-time dynamic and realized the monitoring of the oxidation level of rapeseed oil. This study provided technical support for quality control in the processing of rapeseed oil.
光谱数据融合技术是光谱分析领域的关键创新,是连接不同光谱信息的桥梁。该技术能够整合来自不同平台的光谱数据,对同一样品批次进行综合分析,充分利用各光谱间的互补性和协同效应,以获得更为全面和科学的样品特征信息。这不仅增强了光谱技术在实际应用中的适用性和准确性,也有效克服了单一光谱信息的局限性[15]。根据数据特性和分析需求,融合策略可细分为低层融合(Low-level data fusion LLDF)、中层融合(Mid-level data fusion,MLDF)、高层融合(High-level data fusion,HLDF)以及多区块融合,正确选择融合策略对提升模型的泛化能力和降低过拟合风险至关重要[16]。该技术已广泛应用于食品成分分析、物种鉴定、真伪鉴别以及产地追溯等领域[17-18]。然而,现有文献在光谱数据融合技术的应用上存在一定的局限性。例如,An等[19]采用集成近红外、中红外和拉曼光谱,对比3种数据融合技术,对馒头老化程度进行定量分析,结果表明HLDF在评估馒头的可溶性直链淀粉含量、结晶程度及硬度方面表现最为出色,其中对可溶性直链淀粉含量和结晶程度及硬度的R²均达到 0.9以上。Xu等[20]利用近红外和拉曼光谱的数据融合技术鉴别鳕鱼真伪,并对比单一光谱分析和3种融合策略,结果表明HLDF为最优策略,实现了98%的准确率和92%的敏感度。尽管这2项研究突显了HLDF在食品成分分析和真伪鉴别中的显著优势,但该方法在处理过程中存在重要信息丢失问题,因此未能充分体现光谱数据间的互补性。
在建立预测模型之前,采用Kennard-Stone算法以7∶3的比例将样本划分为训练集和测试集。利用偏最小二乘(Partial least squares regression,PLSR)建立菜籽油氧化指标定量模型,PLSR诠释了多元线性回归、典型相关性分析和主成分分析相结合的优势,同时也是光谱中多元校正分析中应用最广泛的建模方法[36]。
ChangJ L. Research on influencing factors of consumer's consumption behavior for conventional processed rapeseed oil in chengdu[D].Chendu: Southwestern University of Finance and Economics, 2019 (in Chinese)
[3]
CuiY, HaoP, LiuB, MengX. Effect of traditional Chinese cooking methods on fatty acid profiles of vegetable oils.Food Chemistry,233:77-84
[4]
ShahidiF, ZhongY. Lipid oxidation and improving the oxidative stability[J]. Chemical Society Reviews, 2010, 39(11): 4067-4079
WangY P, YaoX H, CaoY Q, ChangJ, RenH D, ZhangC C, WangK L. Study on the quality and oxidation stability of the oils from several oil plants[J/OL].China oil and fats:1-14. --. in Chinese)
[8]
ZhangN, LiY L, WenS S, SunY W, ChenJ, GaoY, SagymbekA, YuX Z. Analytical methods for determining the peroxide value of edible oils: A mini-review [J]. Food Chemistry, 2021, 358: 129834
[9]
HoriK, KohF H, TsumuraK. A metabolomics approach using LC TOF-MS to evaluate oxidation levels of edible oils[J]. Food Analytical Methods, 2019, 12(8): 1799-1804
[10]
VandemoorteleA, HeynderickxP M, LeloupL, De MeulenaerB. Kinetic modeling of malondialdehyde reactivity in oil to simulate actual malondialdehyde formation upon lipid oxidation[J]. Food Research International, 2021,140: 110063
[11]
LiJ W, SunX Y, LiuY F. Analysis and detection of edible oil oxidation[J]. Lipid Technology, 2016, 28(8-9): 145-148
[12]
HussainN, SunD W, PuH B. Classical and emerging non-destructive technologies for safety and quality evaluation of cereals: A review of recent applications [J]. Trends in Food Science & Technology, 2019, 91: 598-608
[13]
ZareefM, ChenQ S, HassanM M, ArslanM, HashimM M, AhmadW, KutsanedzieF Y H, AgyekumA A. An overview on the applications of typical non-linear algorithms coupled with NIR spectroscopy in food analysis[J]. Food Engineering Reviews, 2020, 12(2): 173-190
[14]
ChenJ L, LinH, CaoL M, SuiJ X, WangL, FangX, WangK Q. On-site detection of chloramphenicol in fish using SERS-based magnetic aptasensor coupled with a handheld Raman spectrometer [J]. Spectrochimica Acta Part A, Molecular and Biomolecular Spectroscopy, 2023, 303: 123211
GuoJ, LiK K. Advancements in Adulteration Detection of Edible Oils via Infrared Spectroscopy and Raman Spectroscopy Techniques: A Comprehensive Review.[J/OL].Journal of the Chinese Cereals and Oils Association,1-19. in Chinese)
[17]
WangX, EsquerreC, DowneyG, HenihanL, O’CallaghanD, O’DonnellC. Development of chemometric models using Vis-NIR and Raman spectral data fusion for assessment of infant formula storage temperature and time[J]. Innovative Food Science & Emerging Technologies, 2021, 67: 102551
[18]
CocchiM. Data fusion methods and applications. [M]. Elsevier, 2019:253-260
[19]
SorberL, Van BarelM, De LathauwerL. Structured data fusion[J]. IEEE Journal of Selected Topics in Signal Processing, 2015, 9(4): 586-600
[20]
ZhaoJ, TianG, QiuY Y, QuH B. Rapid quantification of active pharmaceutical ingredient for sugar-free Yangwei Granules in commercial production using FT-NIR spectroscopy based on machine learning techniques [J]. Spectrochimica Acta Part A, Molecular and Biomolecular Spectroscopy, 2021, 245: 118878
[21]
SongY, WangX Z, XieH L, LiL Q, NingJ M, ZhangZ Z. Quality evaluation of Keemun black tea by fusing data obtained from near-infrared reflectance spectroscopy and computer vision sensors[J]. Spectrochimica Acta Part A, Molecular and Biomolecular Spectroscopy, 2021, 252: 119522
[22]
AnH J, ZhaiC, ZhangF, MaQ Y, SunJ F, TangY W, WangW X. Quantitative analysis of Chinese steamed bread staling using NIR, MIR, and Raman spectral data fusion [J]. Food Chemistry, 2023, 405(Pt A): 134821
[23]
XuY, KoidisA, TianX G, XuS, XuX Y, WeiX Q, JiangA M, LeiH T. Bayesian fusion model enhanced codfish classification using near infrared and Raman spectrum[J]. Foods, 2022, 11(24): 4100
NiuF H, LiangJ M, ZhangY Q, JiangY R, WangM J, ShangP L, Preliminary study on the oxidation mechanism of lipid under the accelerated conditions of o the OSI method[J]. Journal of the Chinese Cereals and Oils Association, 2014, 29(10): 67-71 (in Chinese)
GB/T 5009.37—2003 Method for analysis of hygienic standard of edible oils[S]. Beijing: Beijing: Standards Press of China, 2004 (in Chinese)
[28]
GB/T 24304—2009动植物油脂 茴香胺值的测定 [S]. 北京:中国标准出版社,2010
[29]
GB/T 24304—2009—Determination of anisidine value Animal and vegetable fats and oils[S]. Beijing: Beijing: Standards Press of China, 2010 (in Chinese)
[30]
GB/T 22500—2008动植物油脂 紫外吸光度的测定[S]. 北京:中国标准出版社,2008
[31]
GB/T 22500—2008:Determination of ultraviolet absorbance expressed as specific UV extinctionAnimal and vegetable fats and oils [S]. Beijing: Beijing: Standards Press of China, 2009 (in Chinese)
[32]
王凯. 迷迭香提取物与茶多酚棕榈酸酯协同抗油脂氧化及机理解析[D].江南大学,2021
[33]
WangK. Study of synergistic interactions between rosemary extract and tea polyphenol palmitate on oil[D]. Jiangnan University,2021 (in Chinese)
TangY Y, lvH Y, WangW N, ShiL Y, GuangP Y, TangZ H, FanW. Rapid identification of Camellia oil authenticity based on spectral fusion[J]. Journal of Food Safety & Quality, 2023, 14(20): 33-45 (in Chinese)
[36]
LanZ W, ZhangY, SunY, JiD, WangS M, LuT L, CaoH, MengJ. A mid-level data fusion approach for evaluating the internal and external changes determined by FT-NIR, electronic nose and colorimeter in Curcumae Rhizoma processing [J]. Journal of Pharmaceutical and Biomedical Analysis, 2020, 188: 113387
YangH X, TangX P, YangZ M, ZhangJ, LuY J, WuW D. Application research and prospects of multispectral technology in non-destructive testing of food[J]. Science and Technology of Food Industry, 2024, 45(8): 350-357 (in Chinese)
[39]
GeurtsB P, EngelJ, RafiiB, BlanchetL, SuppersA, SzymańskaE, JansenJ J, BuydensL M C. Improving high-dimensional data fusion by exploiting the multivariate advantage[J]. Chemometrics and Intelligent Laboratory Systems, 2016, 156: 231-240
[40]
WangF X, WangC G. Improved model for starch prediction in potato by the fusion of near-infrared spectral and textural data[J]. Foods, 2022, 11(19): 3133
[41]
LiY, HuangY, XiaJ J, XiongY M, MinS G. Quantitative analysis of honey adulteration by spectrum analysis combined with several high-level data fusion strategies[J]. Vibrational Spectroscopy, 2020, 108: 103060
SunF, ZhanS Y, GuoN, LiangS W, QiaoY J, WangS M. System modeling on Chinese herbal medicine process based on multi-block partial least squares[J]. China Journal of Chinese Materia Medica, 2018, 43(16): 3270-3278 (in Chinese)
[44]
NæsT, TomicO, MevikB H, MartensH. Path modelling by sequential PLS regression[J]. Journal of Chemometrics, 2011, 25(1): 28-40
[45]
Lauzon-GauthierJ, ManolescuP, DuchesneC. The Sequential Multi-block PLS algorithm (SMB-PLS): Comparison of performance and interpretability [J]. Chemometrics and Intelligent Laboratory Systems, 2018, 180: 72-83
[46]
TaoL Y, ViaB, WuY J, XiaoW, LiuX S. NIR and MIR spectral data fusion for rapid detection of Lonicera japonica and Artemisia annua by liquid extraction process[J]. Vibrational Spectroscopy, 2019, 102: 31-38
[47]
DupuyN, GaltierO, OllivierD, VanlootP, ArtaudJ. Comparison between NIR, MIR, concatenated NIR and MIR analysis and hierarchical PLS model. Application to virgin olive oil analysis [J]. Analytica Chimica Acta, 2010, 666(1/2): 23-31
[48]
YuX Z, LiQ H, SunD J, DongX B, WangT. Determination of the peroxide value of edible oils by FTIR spectroscopy using polyethylene films[J]. Analytical Methods, 2015, 7(5): 1727-1731
DongC W, LiangG Z, AnT, WangJ J, ZhuH K. Near-infrared spectroscopy detection model for sensory quality and chemical constituents of black tea[J]. Transactions of the Chinese Society of Agricultural Engineering, 2018, 34(24): 306-313 (in Chinese)
WangZ W, ChiQ, GuoW C, ZhaoC J. Internal quality detection of apples during late developmental period based on near-infrared spectral technology[J]. Transactions of the Chinese Society for Agricultural Machinery, 2018, 49(5): 348-354 (in Chinese)