拉曼光谱对茶油三元体系掺伪检测研究  

Raman spectroscopy detection of camellia oil adulteration in ternary system

作　　者：郭佳[1] 郭郁葱[2] 姜红 李开开[1] GUO Jia;GUO Yucong;JIANG Hong;LI Kaikai(College of Investigation,People’s Public Security University of China,Beijing 102600,China;Beijing National Research Center for Molecular Sciences,Institute of Chemistry,Chinese Academy of Sciences,Beijing 100190,China;Criminal Investigation Department,Gansu Police Vocational College,Lanzhou 730046,China)

机构地区：[1]中国人民公安大学侦查学院,北京102600 [2]中国科学院化学研究所,北京分子科学国家研究中心,北京100190 [3]甘肃警察职业学院刑事侦查系,甘肃兰州730046

出　　处：《食品与发酵工业》2024年第22期327-333,共7页Food and Fermentation Industries

基　　金：国家自然科学基金(22203104,42175133);中国人民公安大学刑事科学技术双一流创新研究专项(2023SYL06)。

摘　　要：该研究采用拉曼光谱技术对茶油三元体系掺伪进行定量检测研究,通过对比不同预处理方法、建模方法及优化算法的优劣,确定最优的大豆油、玉米油、茶油的多元掺伪检测模型。利用一阶微分、二阶微分、多元散射矫正、标准正态变换等不同预处理方法消除外界因素对光谱的影响,竞争性自适应重加权算法提取特征光谱波段,通过偏最小二乘回归和支持向量机建立茶油掺伪检测模型,分别采用网格搜索法和粒子群算法对支持向量机进行优化。基于标准正态变换预处理后所建立模型效果最佳,大豆油和茶油的最佳预测模型为基于粒子群算法优化的支持向量机,玉米油的最佳预测模型为基于网格搜索法优化的支持向量机,大豆油、玉米油和茶油的预测集决定系数R2和预测均方根误差分别为0.9986、0.9994、0.9999和2.73%、1.62%、0.40%。该研究确定了最优的大豆油、玉米油、茶油的多元掺伪检测模型,针对市场茶油的掺伪检测,基于拉曼光谱分析和优化算法的支持向量机模型为茶油的无损快速定量检测提供了一定的参考和借鉴。Camellia oil is often adulterated with other cheap cooking oils in the market.Raman spectroscopy was employed to quantitatively detect adulteration in the ternary system of camellia oil.The optimal multivariate adulteration detection model for soybean oil,corn oil,and camellia oil was determined by comparing the advantages and disadvantages of different preprocessing methods,modelling methods,and optimization algorithms.To eliminate the effect of external factors on the spectrum,four preprocessing methods were employed,including first-order differentiation,second-order differentiation,multiple scattering correction,and standard normal variation.The competitive adaptive reweighted sampling algorithm was employed to extract the characteristic spectral bands.The adulteration detection model for camellia oil was developed using partial least squares regression and support vector machine regression.The support vector machine was optimized using both grid search and particle swarm optimization algorithms.Among the preprocessing methods,the models based on standard normal variation yielded the most favorable outcomes.The particle swarm optimization-support vector machine(PSO-SVM)model proved to be highly effective in predicting soybean oil and camellia oil,while the grid search-support vector machine(GS-SVM)model demonstrated superior performance in predicting corn oil.The predicted coefficient of determination(R 2)and the predicted root mean square error(RMSEP)for soybean oil,corn oil,and camellia oil were 0.9986,0.9994,and 0.9999,and 2.73%,1.62%,and 0.40%,respectively.The study presented an optimal multivariate model for detecting adulteration in soybean oil,corn oil,and camellia oil.The support vector machine model,based on Raman spectral analysis and optimization algorithms,provides a valuable reference for non-destructive and rapid quantitative detection of camellia oil adulteration in the market.

关 键 词：茶油 拉曼光谱 掺伪检测 偏最小二乘回归 粒子群算法优化 支持向量机 

分 类 号：TS227[轻工技术与工程—粮食、油脂及植物蛋白工程] O657.37[轻工技术与工程—食品科学与工程]