摘要
近红外反射光谱技术是一种高效、快速的现代分析技术,应用过程中需要建立相应的数学模型。收集油菜籽样品600余份,采用国际(家)标准方法对芥酸、硫苷、含油率进行测试,根据各模型的要求,选择不同的具有代表性的样品,建立数学模型。内部交叉证实法对模型验证结果显示:高芥酸、低芥酸、硫苷、含油率模型复相关系数分别为0.9677、0.9318、0.9820、0.9626,内部验证均方差(RMSECV)分别为4.41、2.24、4.62、0.543。外部样品检测法分别用已建模型和国际(家)标准方法对油菜籽样品芥酸、硫苷、含油率进行检测,两者检测结果基本一致。综合内部和外部验证结果表明:建立的模型能够满足油菜籽硫甘、芥酸及油份含量快速检测的要求。另对油菜自身含水率对数学模型的影响做了初步探索。
Near-infrared reflectance （NIR） spectroscopy was a modern high-efficient and prompt analysis technique. But the mathematical models must be created before detection. More than 600 rapeseeds samples were collected. Erucic acid, total glucosinolates and oil content of them were detected by international or national standard methods. Different representative samples were chosen to create mathematical models according to different requirements. Cross validation showed that the multiple correlation coefficient of high erucic acid model, low erucic acid model, glucosinolates model and oil model were 0. 9677, 0. 9318, 0. 9820, 0. 9626 respectively. The RMSECV were 4.41, 2.24, 4.62, 0. 543 respectively. The test results of erucic acid, glucosinolates and oil content in outer samples detected by the created models were basically consistent with that detected by the standard methods. Cross validation and outer samples test have proved that the created models were effective and reliable. The impact of water content in rapeseed samples in the models was studied too.
出处
《激光生物学报》
CAS
CSCD
2009年第6期815-818,共4页
Acta Laser Biology Sinica
关键词
近红外光谱
数学模型
油菜
硫苷含量
芥酸含量
油份含量
NIRS
mathematical models
rapeseed
glucosinolates content
erucic acid content
oil content
