食用向日葵粗蛋白含量近红外光谱模型的建立及可行性分析

Development and Reliability of Near Infrared Spectroscopy(NIS) Models of Crude Protein Content in Confectionery Sunflower

以自育的食用向日葵材料为测试对象,用化学方法测定粗蛋白含量,对照近红外漫反射光谱,通过偏最小二乘法分别建立了食用向日葵籽粒、籽仁和籽粒粉末粗蛋白含量的近红外光谱模型。结果表明:籽仁模型和籽粒粉末模型预测未知样品粗蛋白含量的精确度较高,可以用于向日葵粗蛋白含量的定量测定,而籽粒模型与其他2个模型相比,预测精确度相对较低,但由于其不破坏籽粒和种皮的优点,可用于粗蛋白含量的初步评估和高蛋白向日葵材料的筛选。

With the self-bred confectionery sunflower material as the detection objects, the protein content was measured by using chemical methods. In comparison with the near-infrared diffuse reflectance spectroscopy, the near-infrared reflectance spectroscopy（NIRS） models of seed, kernel and crude protein content of seed powder of confectionery sunflower was established by means of partial least-squares（PLS） regression. The results showed that measuring accuracy of kernel model and seed powder model is higher, and they can be available for quantitative determination of confectionery sunflower crude protein content. Measuring accuracy of seed model is lower than that of the other two models, but because the merit of not destroying seeds and husk, it can be used for a preliminary assessment of crude protein content and high protein material screening.