肖朝耿,叶沁,卢文静,谌迪,唐宏刚,杨慧娟,孟祥河,陈黎洪.基于DR-FTIR技术结合biPLS法测定精米粉中蛋白质含量的研究[J].中国粮油学报,2019,34(6):129-133
基于DR-FTIR技术结合biPLS法测定精米粉中蛋白质含量的研究
Rapid Determination of Protein Content in Milled Rice Flour by DR-FTIR Technology Combined with Interval Partial Least Squares
投稿时间:2018-09-04  修订日期:2018-12-06
DOI:
中文关键词:  中红外光谱  蛋白质  精米  区间偏最小二乘法  测定
英文关键词:Diffuse  Reflectance Fourier  transform infrared  spectroscopy, protein, milled  rice flour, back  interval partial  least squares, determination
基金项目:
作者单位邮编
肖朝耿* 浙江省农业科学院食品科学研究所 310021
叶沁 浙江工业大学海洋学院 
卢文静 浙江省农业科学院食品科学研究所 
谌迪 浙江省农业科学院食品科学研究所 
唐宏刚 浙江省农业科学院食品科学研究所 
杨慧娟 浙江省农业科学院食品科学研究所 
孟祥河 浙江工业大学海洋学院 
陈黎洪 浙江省农业科学院食品科学研究所 
摘要点击次数: 839
全文下载次数: 36
中文摘要:
      开发了傅里叶中红外漫反射光谱技术(DR-FTIR)结合向后区间偏最小二乘法(biPLS)快速分析精米中蛋白质含量的分析方法。通过波谱预处理、优化建模区间,系统地比较了DR-FTIR及近红外漫反射红外光谱(DR-NIR)的模型效果。研究结果表明,DR-FTIR、DR-NIR均能快速准确的测定精米中蛋白质含量,但DR-FTIR法灵敏度优于DR-NIR法。BiPLS区间选择结果显示,以1700-1400, 1300-1200, 900-600 cm-1波段建模,模型效果最优,其相关系数R为0.9901,RMSEC和RMSEP分别为0.13、0.25,Y预测 = 0.99X实际 + 0.0082,R = 0.9923,预测值与实际值高度相关。10组外部验证试验相对误差为1.72%,表明预测精确度较高、模型稳定性好,说明DR-FTIR所开发的精米中蛋白质含量的分析模型是可行的。
英文摘要:
      The analysis method for rapidly determining protein content in milled rice flour by Diffuse Reflectance Fourier transform infrared spectroscopy (DR-FTIR) combined with back interval partial least squares (biPLS) was studied. The effects of DR-FTIR and DR-near infrared spectroscopy (NIR) models were systematically compared by spectral pretreatment and optimization of spectral modeling interval. The results showed that both DR-FTIR and DR-NIR were effective in term of rapid determination of the protein content in milled rice flour. However, DR-FTIR method had higher sensitivity than DR-NIR method. The results of biPLS interval selection demonstrated that the best model was obtained based on the 1700-1400, 1300-1200, 900-600 cm<sub>-1</sub>bands, coefficient of correlation R was 0.9901, RMSEC was 0.13, and RMSEP was 0.25. The DR-FTIR predictive values were highly correlated with the actual values with a regression equation of Y<sub>prediction </sub>= 0.99X<sub>actuality </sub>+ 0.0082,R = 0.9923. The relative standard deviation of 10 groups of external validation tests (1.72%) showed that the model had higher prediction accuracy, better stability, and it demonstrate that the analysis model of protein content in milled rice flour developed by DR-FTIR was feasible.
关闭