傅立叶变换近红外光谱技术分析完整油菜籽含油量数学模型的建立

Establishment of math model of Fourier transform NIRS analysis for oil content in intact rapeseed

利用傅立叶近红外技术和大量具有代表性油菜籽样品建立含油量测定模型，内部交叉法和外部检验集样品对模型检验结果：含油量模型内部交叉验证均方差为O．755，最佳主成分维数为9，采用一阶导数＋矢量归一化方法预处理，谱区为7502．2—6098．2cm^-1、4601．6～4246．7cm^-1时，所建模型对油菜籽含油量测定结果与标准方法测定结果基本一致。说明建立的近红外分析模型能够满足油菜育种中对初级世代材料含油量的快速测定筛选要求。

The model for determination of oil content red set up by near- infrared reflectance spectroscopy technology, and the model was tested by using of cross validation and test sample determination. The resuits showed that the RMSECV and optimal ranks of the model were 0.775 and 9 respectively. Comparison of the oil content value by NIRS and the standard methods showed good correlation when the optimal model was developed by spectral data pretreatment of first derivative ＋ vector normliztion in 7 502.2 - 6 098.2 cm^-1 and 4 601.6 -4 246.7 cm^-1 .These results indicated that the method of NIKS established in present study had relatively high accuracy in prediction test for oil content in primary generation of rapeseed breeding.