近红外方法预测馒头品质

本研究利用近红外方法预测馒头品质评分参数,并取得了较好的预测效果。从150份小麦粉样品中筛选出31个具有不同品质特性的小麦粉样品,使用FOSS Infraxact Lab近红外光谱仪在570~1 850 nm波长下扫描,按实验室馒头制作方法制作馒头并进行评分,使用WinISIⅢ处理软件处理数据,结合修正偏最小二乘法(MPLS)建立了定标模型,高径比、比容、色泽、外观性状等11项参数的定标决定系数(r2)在0.60~0.94之间,定标标准误差(SEC)范围为0.02~2.06,并且取得较好的交叉验证相关系数(1-VR)和较低的交叉验证标准误差(SECV),结果表明利用近红外方法预测馒头品质评分参数具有可行性。

Assessment of Mantou quality by Near Infrared Reflectance method

In this study, the near infrared method was firstly used to predict Mantou quality scoring parameters, and achieved gold prediction results. Thirty-one samples with different quality characteristics were selected from 150 flours. The near infrared spectrum of the thirty one samples were obtained by using the FOSS near-infrared spectroscopy (Infraxact Lab) at wavelengths between 570~1850 nm, and Mantou made from each sample was scored. The software WinISI III combined with modified partial least squares was applied to build calibration models. The ranges of Regression squared and Standard error of calibration of eleven indicators, including the height-diameter ratio, specific volume, color, appearance, surface structure, etc. were 0.60~0.94 and 0.02~2.06 respectively. Each parameter exhibited a high cross-validation relation coefficient (1-VR) value and relative low standard error of cross validation (SECV) value. These results suggested that predicting score of Mantou using the NIR models was feasible.