用近红外反射光谱法分析稻米3种必需氨基酸含量的研究  

Analysis of Three Types of Essential Amino Acids Content in Milled Rice by Near Infrared Reflectance Spectroscopy

作　　者：吴建国[1] 石春海[1] 张小明[1] 费万辛[1] 

机构地区：[1]浙江大学农业生物技术学院农学系,浙江杭州310029

出　　处：《作物学报》2003年第5期688-692,共5页Acta Agronomica Sinica

基　　金：浙江省科技厅项目 (0 1110 2 471);浙江省"15 1人才基金"资助

摘　　要：采用近红外反射光谱分析技术来测定水稻精米粉样的 3种必需氨基酸含量 ,揭示不同数学处理方法、光谱散射校正和回归统计方法对建立回归方程的影响。研究表明 ,光谱数据的数学预处理对发展定标回归方程有很大影响 ,对于不同的氨基酸 ,通过比较均可找到一些定标效果较理想的数学处理组合 ,如“2 ,3,3,1”等。光谱散射处理对氨基酸的定标改善也有一定作用。除趋势变换法外 ,标准正态变量转换、标准正态变量转换 +趋势变换法、乘性散射校正、加权乘性散射校正和逆向乘性散射校正这 5种校正均有较为一致的处理效果。回归统计方法 ,以PLS和MPLS建立的回归方程效果好。通过以上三个方面的调试 ,可以找到一些效果较好的组合 ,使得 3种必需氨基酸 :苏氨酸、胱氨酸、赖氨酸的校正决定系数 (RSQ1)高达 0 92 7、0 888和 0 82 3,而定标工作标准差 [SEP(C) ]分别小至 0 2 5 8、0 179和 0 2 98。因此 ,近红外分析技术可用于稻米的苏氨酸、胱氨酸和赖氨酸 3种必须氨基酸的测定。By using milled rice as test samples, the methods of mathematics pretreatment, scatter correction and different regression methods were compared to establish a better regression equation for three kinds of essential amino acid by near infrared reflectance spectroscopy (NIRS). Some good methods of mathematics pretreatment, such as “2,3,3,1”, could be selected from those that had some effects on NIRS calibration for different amino acid. Scatter correction methods of NIRS were benefit to develop good calibration equations for theses essential amino acids. Except “Detrend (D)”, the other five methods of scatter correction, “Transformation of Standard Normal Variate (SNV)”, “SNV+D”, “Standard Multiplicative Scatter Corrections (Standard MSC)” or “Inverse Multiplicative Scatter Corrections (Inverse MSC)” could be used in option for their consistent effects on calibration. Regression method had obvious effect on calibration with two better methods, “Modified Partial Least Square (MPLS)” and “Partial Least Square (PLS)”. Based on the three aspects, mathematic treatment, scatter correction and regression method, suitable regression equations for essential amino acid (thronine, cysteine and lysine), could be screened. The ideal calibration coefficient (RSQ1) for thronine, cysteine and lysine, were 0.927, 0.888 and 0 823, respectively, and the correction standard errors of performance (SPE(C)) were 0.258, 0.179 and 0 298, respectively. Therefore, the NIRS could be introduced to estimate essential amino acids in ground rice. The NIRS application in milled rice samples for nutrition quality evaluation in breeding programs was discussed.

关 键 词：近红外反射光谱法 稻米 氨基酸 含量 精米粉样 数学处理 光谱散射校正 回归方程 

分 类 号：S511[农业科学—作物学]