稻米品质近红外光谱法分析中定标样品选择与模型优化.PDFVIP

核 农 学 报 　2008 ,22 (6) :823～828 823 　 J ournal of N uclear Agricultural Sciences ( ) 文章编号 2008 0682306 稻米品质近红外光谱法分析中定标样品选择与模型优化 陆艳婷 　张小明　叶胜海 　祁永斌 　金庆生 (浙江省农业科学院作物与核技术利用研究所 ,杭州 浙江 　310021) 摘 　要 :通过对粳型水稻精米粉的直链淀粉含量和氨基酸含量近红外光谱检测模型的定标样品数量和优 化的分析 ,用新复极差检验对不同样品组的粳稻精米粉直链淀粉含量和谷氨酸 、亮氨酸含量的近红外建 ( 2 ) ( ) 模效果校正决定系数 R 和交叉检验均方误差 RMSECV 进行了评价 ,发现样品数为 2 16 份 、114 份 、81 份的直链淀粉含量样品组间无显著差异 ,200 份 、132 份 、91 份谷氨酸和 200 份 、113 份和 70 份亮氨酸含 量的样品组间无显著差异 。结果表明 ,优选合适的样品及数量 ,并对模型进行优化 ,不仅可保持定标方 程的精度 ,也可节约成本 、减少工时 。在本试验中 ,直链淀粉 、谷氨酸和亮氨酸含量分别选择 81 、91 、70 份分布均匀的样品 ,可满足各自的建模要求 。 关键词 :近红外光谱法 ;水稻 ;样品数量 ;模型优化 THE CALIBRATION SAMPL E SEL ECTION AND MODEL OPTIMIZATION IN RICE QUALITY ANALYSIS BY NEAR INFRARED SPECTROSCOPY LU Yanting 　ZHAN G Xiaoming 　YE Shenghai 　QI Yongbin 　J IN Qingsheng ( Institute of Crop and N uclear Technology Utilization , Zhej iang Academy of Agricultural Sciences , Hangzhou , Zhej iang 　310021) Abstract :The number of calibration sample and model optimization of using nearinfrared spectroscopy for the measurement of amylose content and amino acid contents of milled j ap onica rice powder were studied . The Duncan ’s multiple ranger test was done for different number sample set of rice amylose , glutamic acid and leucine contents determination through nearinfrared model predication analysis , which was appreciated by the optimal calibration determination coefficient ( R2 ) and cross examination mean square errors (RMSECV) . The test results showed that there were no significant difference of the amy