论文部分内容阅读
文章以66个小麦样品为实验材料,其中33个为建模集,剩余33个为预测集,利用广义逆矩阵直 接确定傅里叶变换近红外全谱分析回归模型中的回归系数,建立了用于蛋白质定量分析的近红外全谱回归 模型。用此模型对预测集中的样品进行预测,结果与凯氏定氮法测定结果间的相关系数为r=0.979 9,平均 相对误差为1.76%,表明由广义逆矩阵方法所建近红外全谱定量分析回归模型有较好的分析结果。所建模 型不仅可用于对样品的实际分析,而且可根据回归模型中各个系数了解各个波长点处的光谱信息对模型预 测值的贡献,从而可理解并解释傅里叶变换近红外全谱回归模型的物理学与化学意义。
In this paper, 66 wheat samples were selected as experimental materials, of which 33 were modeled sets and the remaining 33 were prediction sets. The regression coefficients of Fourier transform NIRS were directly determined by generalized inverse matrix. Near Infrared Full Spectrum Regression Model for Protein Quantitative Analysis. The model was used to predict the sample in the prediction set. The correlation coefficient between the results and the Kjeldahl method was r = 0.979 9 and the average relative error was 1.76%, which indicated that the model was constructed by the generalized inverse matrix method IR full-spectrum quantitative regression model has a good analysis results. The model can be used not only for the actual analysis of the sample, but also for each coefficient in the regression model to understand the spectral information at each wavelength point of the contribution to the model predictions, which can understand and explain the Fourier transform near-infrared full-spectrum regression model The physical and chemical significance.