Quantitative Analysis of Deep-Frying Oil Adulterated Virgin Olive Oil Using Vis-NIR Spectroscopy with iPLS
XIAN Rui-yi1, HUANG Fu-rong1*, LI Yuan-peng1, PAN Sha-sha1, CHEN Zhe1, CHEN Zhen-qiang1, WANG Yong2
1. Opto-electronic Department of Jinan University，Guangzhou 510632，China 2. Food Science and Engineering Department of Jinan University，Guangzhou 510632，China
摘要： 为探寻一种快速可靠的分析方法用于橄榄油中掺杂煎炸老油含量的测定，实验采用可见和近红外透射光谱分析技术结合区间偏最小二乘法(interval partial least squares, iPLS)、联合区间偏最小二乘法(synergy interval partial least squares, SiPLS)和反向区间偏最小二乘法(backward interval partial least squares, BiPLS)，对掺杂不同含量煎炸老油的橄榄油建模分析，并对不同模型比较优选。采集样品400~2500 nm范围内的光谱，对光谱数据进行Savitzky-Golay(SG)平滑去噪。剔除奇异样本后，采用sample set partitioning based on joint X-Y distance(SPXY)法划分样本集，以不同的iPLS优选建模区域，建立煎炸老油含量预测模型。结果表明：对掺杂不同含量煎炸大豆油的橄榄油，采用划分20个区间，选择2个子区间[4, 16]建立的SiPLS模型预测效果最好，相关系数(Rp)达0.998 9，预测均方根误差(RMSEP)为0.019 2。对掺杂不同含量煎炸花生油的橄榄油，采用划分20个区间，选择2个子区间[2, 16]组合建立的SiPLS和BiPLS模型具有相同的预测效果，预测均方根误差(RMSEP)为0.0120，均优于iPLS模型。此外，与SiPLS模型相比，BiPLS模型运算量少，速度快。由此可见，基于掺杂油样品的可见和近红外透射光谱，分别采用组合区间偏最小二乘法(SiPLS)和反向区间偏最小二乘法(BiPLS)优选建模光谱区域，可以对橄榄油中掺杂煎炸大豆油和煎炸花生油含量进行准确测定。而且，实验过程无需对掺杂油样品进行预处理，无环境污染，操作简单，快速无损。
关键词：可见和近红外透射光谱;区间偏最小二乘法;掺伪;煎炸老油;定量分析
Abstract：To explore a rapid and reliable method for quantitative analysis of deep-frying oil adulterated virgin olive oil, visible and near infrared(Vis-NIR) spectroscopy and three improved partial least squares methods, including interval Partial Least Squares (iPLS), synergy interval partial least squares (SiPLS) and backward interval partial least squares (BiPLS) were employed to establish predicting models of doping content based on virgin olive oil adulterating different levels and different types of deep-frying oil. And the models were compared in order to choose the best one. The Vis-NIR spectroscopy ranged from 400 to 2 500 nm was obtained directly from the adulterated samples, and the spectroscopic data was preprocessed with Savitzky-Golay (SG). Then, the samples were divided into calibration set and test set by Sample Set Partitioning based on Joint X-Y Distance (SPXY) after rejecting the odd samples. At last, the predicting models of doping content were built by using different interval partial least squares methods. The results showed that the optimal model for predicting the doping content of deep-frying soybean oil in virgin olive oil was obtained with SiPLS method that separated the whole spectra into 20 intervals and combined the fourth and the sixteenth intervals. The SiPLS model had correlation coefficient (r) of 0.998 9 and root mean standard error of prediction (RMSEP) of 0.019 2. In addition, for deep-frying peanut oil adulterated virgin olive oil, the SiPLS and BiPLS models with interval 2 and interval 16 which the whole spectra was separated into 20 intervals, had same results. The RMSEP was 0.012 0, lower than iPLS model. Moreover, compared to SiPLS method, BiPLS method saved computation and was more efficient. Overall, through selecting the effective wavelength range, SiPLS method and BiPLS method could accurately predict the doping content of deep-frying oil in virgin olive oil based on its’ Vis-NIR spectroscopy. In addition, this fast and nondestructive experiment doesn’t need sample pretreatment with advantages of no environment pollution, easy operation.
Key words：Vis-NIR spectroscopy;Interval partial least squares regression (iPLS);Adulteration;Deep-frying oil;Quantitative analysis
[1] Clark W L, Serbia G W. Food Technology, 1991, 45(2): 84. [2] Freire P C M, Lobo L C B, Freitas G D, et all. Food Science and Technology, 2013, 33(3): 569. [3] Weisshaar Rüdiger. European Journal of Lipid Science & Technology, 2014, 116(6): 716. [4] Sebastian A, Ghazani S M, Marangoni A G. Food Research International, 2014, 64: 420. [5] Torrecilla J S, Cancilla J C. Journal of Food Engineering, 2013, 118(4): 400. [6] Wrona M, Pezo D, Nerin C. Food Chemistry, 2013, 141(4): 3993. [7] Ni Yongnian, Mei Minghua, Serge Kokot. Chemometrics and Intelligent Laboratory Systems, 2011, 105(2): 147. [8] Galtier O, Abbas O, Le Dréau Y, et al. Vibrational Spectroscopy, 2011, 55(1): 132. [9] Rohman Abdul, Nugroho Ardi, Lukitaningsih Endang, et al. Applied Spectroscopy Reviews, 2014, 49(8): 603. [10] Alberto Villar, Santiago Fernández, Eneko Gorritxategi, et al. Chemometrics and Intelligent Laboratory Systems, 2014, 130(2): 68. [11] Julia Kuligowski, David Carrión, Guillermo Quintás, et al. Food Chemistry, 2012, 131(1): 353. [12] Niu Xiaoying, Zhao Zhilei, Jia Kejun, et al. Food Chemistry, 2012, 133(2): 592. [13] Roberto Kawakami Harrop Galvo, Mário César Ugulino Araujo, Gledson Emídio José, et al. Talanta, 2005, 67(4): 736. [14] Nrgaard R, Saudland A, Wagner J, et al. Applied Spectroscopy, 2000, 54(3): 413. [15] Workman Jerry, Weyer Jr Lois. Practical Guide to Interpretive Near-Infrared Spectroscopy(近红外光谱解析实用指南). Translated by CHU Xiao-li，XU Yu-peng，TIAN Gao-you(褚小立，许育鹏，田高友，译). Beijing: Chemical Industry Press(北京：化学工业出版社), 2009. 221.
