摘要
采用近红外光谱(NIRS)分析技术和化学计量方法建立稻谷脂肪酸值的近红外分析模型,并对模型进行预测准确性评价;在建立定标模型的过程中,探讨光谱散射处理、数学(导数)处理等优化处理对定标模型的影响。结果表明:修正偏最小二乘法是建立稻谷脂肪酸值测定定标模型的最佳回归方法,所建立模型的定标相关系数(RSQ)为0.961,定标标准偏差(SEC)为1.9205;内部交互验证相关系数(1-VR)为0.9474,内部交互验证标准偏差(SECV)为2.2511。外部验证的相关系数(r)为0.951,外部验证标准偏差(SEP)为1.934。标准方法与NIRS测定方法测定的稻谷脂肪酸值含量之间的t检验值为1.403,显示两种方法测定结果无显著性差异(P<0.1),预测值与实测值的平均绝对偏差为0.27,说明所建立的稻谷脂肪酸值的NIRS数学模型预测准确性较好,可用于稻谷脂肪酸值的快速预测。
The mathematic models for the prediction of fatty acid content of rice was established with the technique of nearinfrared spectroscopy (NIRS). The result showed that the calibration models developed by the partial least square (PLS) regression were optimum. The statistical values of calibration equation were as follows:the coefficient of correlation (RSQ) of 0.961, the standard error of calibration (SEC) of 1.9205, the determination coefficient of cross-validation (1-VR) of 0.9474, the standard error of cross-validation (SECV) of 2.2511, Regression squared (r) of 0.951, square error of prediction (SEP) of 1.934. The t test value between the chemical standard methods and NIRS method was 1.403 (P〈0.1), suggesting no significant difference between these two methods. The absolute average deviation was 0.27. This NIRS method could be applied to predict the fatty acid content in rice.
出处
《食品科学》
EI
CAS
CSCD
北大核心
2009年第24期347-350,共4页
Food Science
作者简介
范维燕(1982-),女,硕士研究生,主要从事粮食、油脂及植物蛋白工程研究。E-mail:ziruibaihe82@163.com;
通讯作者:林家永(1960-),男,副研究员,本科,主要从事粮食品质与增值利用研究。E—mail:linjy@chinagrain.org