近红外(NIRS)技术分析小样品油菜籽芥酸和含油量的应用研究

以完整油菜籽为样品,采用旋转杯和安培瓶两种样品杯、每种样品分为4×2种不同样品量并通过不同光谱预处理来优化油菜籽芥酸和含油量的近红外分析模型。结果表明:油菜籽各小样品含油量模型的决定系数(R2)从93.93%到96.93%不等,均方差(RMSECV)从0.56到0.79不等;油菜籽各小样品芥酸模型的决定系数(R2)从96.91%到98.42%不等,均方差(RMSECV)从1.73到2.43不等。随着样品量的逐渐增加,油菜籽芥酸和含油量不同样品杯模型各参数逐渐有所优化;同一样品厚度时,油菜小样品芥酸和含油量的旋转样品杯模型各参数均略优于安培瓶样品模型;不同样品量的NIRS模型,W3和W4差异不大,依次优于W2和W1。最小样品量AW1为0.3g。优化油菜小样品模型时,应该选择全部的预处理方法,根据优化结果选择最佳模型。外部检验结果表明:不同重量小样品(W1/0.3g、W2/1.0g、W3/2.0g和W4/4.0g)模型之间及其与标准化学值之间在0.01水平上差异不显著,说明W1和W2小样品模型同样可应用于油菜品质育种材料的分析选择。

Analyses of Oil Content and Erucic Acid in Small Sample of Rapeseed by Near Infrared Reflectance Spectroscopy

Four different mass(W1/0.3 g,W2/1.0 g,W3/2.0 g and W4/4.0 g) of rapeseed samples placed in two types of sample cups,namely rotary cup and ampoule,respectively,were processed by different spectral pretreatments to optimize the mathematical models for the analyses of erucic acid and oil content in small sample of intact rapeseeds by near infrared reflectance spectroscopy(NIRS).The results showed that the determination coefficients of the models for oil and erucic acid content in different small-sample rapeseed were 93.93%-96.93% and 96.91%-98.42%,respectively;and their corresponding mean square deviations were 0.56-0.79 and 1.73-2.43,respectively.The determination coefficient and the mean square deviation of the models were optimized with the gradual increase of sample mass.For the same sample mass,the parameters of NIRS models for rotary cup surpassed those for ampoule.It was found that the least sample mass,AW1 was 0.3 g.When optimizing the model for the small-sample rapeseed,all of the spectral pretreatment methods should be selected.The best model was then selected according to the optimizing result.The external examination result showed that the difference between the models with different mass of small sample(W1/0.3 g,W2/1.0 g,W3/2.0 g and W4/4.0 g) and the difference between the predicted value and the true value obtained by the standard method was insignificant,indicating that the W1 and W2 small sample rapeseed model could be used for the quantitative analysis of erucic acid and oil content in rapeseed breeding and quality selection.