基于近红外光谱对淀粉类的定性及定量分析-《中国调味品》

文章信息/Info

Title:: Qualitative identification and quantitative analysis of starch based on near infrared spectroscopy

摘要:: 淀粉在调味品生产中应用广泛，同时本身也作为调味品具有勾芡等用途。不同种类淀粉分子结构不同导致性质用途不同，本文基于近红外漫反射技术，建立了不同种类淀粉快速无损定性及定量检测方法。通过对73份不同种类的淀粉进行光谱采集，结合主成分分析法（PCA），建立了不同种类淀粉的定性判别模型；基于淀粉种类判别结果，以不同比例的马铃薯和小麦淀粉的混合物为检测对象，采集光谱信息，基于偏最小二乘法（PLS）建立近红外定量分析模型，并对该模型进行了验证，校正模型和预测模型的决定系数分别为0.9988和0.9982，均方根误差分别为0.385和0.347、预测相对分析误差（RPD）为14.3。结果表明，近红外光谱技术可用于淀粉的定量预测，以及不同种类淀粉的快速定性识别。

Abstract:: Starch is widely used in the seasoning industry, and it is also used as a seasoning for hooking. Different kinds of starch have different molecular structures, which lead to different properties. Based on near-infrared diffuse reflection technology, this paper established a rapid and non-destructive qualitative and quantitative detection method for different kinds of starches. A qualitative identification model of different types of starch was established by spectral collection of 73 different types of starch, combined with principal component analysis (PCA). Based on the determination of starch type, a mixture of potato and wheat starch was tested at different ratios. The object, the spectral information was collected, and the near-infrared quantitative analysis model was established based on partial least squares (PLS), and the model was verified. The correlation coefficients of the calibration model and the prediction model were 0.9988 and 0.9982, respectively, and the root mean square error is 0.385 and 0.347, the predicted relative analysis error (RPD) is 14.3. The results show that near-infrared spectroscopy can be used for quantitative prediction of starch and rapid qualitative identification of different kinds of starch.