应用近红外透射光谱技术,采用改进的偏最小二乘法(MPLS)建立糙米蛋白质含量(PC)定量分析数学模型。糙米蛋白质含量预测数学模型校正标准误差(SEC)、交叉检验标准误差(SECV)分别为0.2114、0.2365,校正相关系数(RSQ)和交叉验证相关系数(1-VR)分别为0.9807、0.9768,预测误差与常规分析方法的误差接近。内部交叉检验和外部验证结果表明,近红外定量分析有很高的准确度,近红外光谱法完全可以替代常规分析方法。 Near infrared transmission spectroscopy was used to establish quantitative mathematical model of protein content (PC) in brown rice by using improved partial least squares (MPLS). The SEC and SECV of the mathematic model for prediction of brown rice protein content were 0.2114 and 0.2365, respectively, and the correlation coefficients (RSQ) and cross validation (1-VR) of brown rice were 0.9807 and 0.9768, respectively. The error is close to that of the conventional analysis. Internal cross-validation and external validation results show that NIR quantitative analysis has high accuracy and NIRS can completely replace conventional analytical methods.