摘要
研究采集不同种类、不同氧化程度的84个食用油样本拉曼光谱图,建立关于其过氧化值的偏最小二乘定量检测模型,进行波长筛选,并用“一阶导数法+SNV”对光谱进行预处理,建模相关系数达到0.9528,校正标准偏差为0.5720;预测相关系数为0.9219,相对预测均方差为0.698。结果表明,经光谱预处理和波长筛选优化后,模型准确性和稳健性可靠,误差相对较小,最终得出了合理的食用油过氧化值定量模型。由此可见,研究所建定量模型稳健性良好,利用拉曼光谱分析技术对油品过氧化值进行快速检测具有可行性,并且这种技术有望用于日常食用油品质的快速检测。
Raman spectra of 84 edible oil samples with different types and degrees of oxidation were collected,and a partial least squares quantitative detection model for their peroxide values was established.The wavelength ranges of the spectra were screened,and the spectra were preprocessed by the method of"first derivative+SNV".The modeling correlation coefficient was 0.9528,the root mean square error of calibration was 0.5720,the prediction correlation coefficient was 0.9219,and the root mean square error of prediction was 0.698.The results showed that after spectral preprocessing and wavelength screening optimization,the accuracy and robustness of the model were reliable,and the error was relatively small.Finally,a reasonable quantitative model of edible oil peroxide value was obtained.Therefore,the established quantitative model in the study was robust,and it was feasible to use Raman spectroscopy to quickly inspect the peroxide value of oil.It was expected to be used for the rapid detection of daily edible oil quality.
作者
杜馨
孙晓荣
刘翠玲
李敬琪
李睿雯
DU Xin;SUN Xiaorong;LIU Cuiling;LI Jingqi;LI Ruiwen(Beijing Key Laboratory of Big Data Technology for Food Safety,School of Computer and Information Engineering,Beijing Technology and Business University,Beijing 100048,China)
出处
《中国酿造》
CAS
北大核心
2019年第12期171-174,共4页
China Brewing
基金
北京市教委科技计划一般项目(KM201810011006)
北京市自然科学基金项目(4182017)
全国大学生科学研究与创业行动计划(201810011090)
关键词
食用油
过氧化值
拉曼光谱
偏最小二乘法
定量分析
edible oil
peroxide value
Raman spectroscopy
partial least squares
quantitative analysis
