论文部分内容阅读
利用近红外光谱结合偏最小二乘法建立了快速测定益母草颗粒的盐酸水苏碱的定量分析模型,模型的决定系数为0.9622,内部交叉验证均方差为0.0576;然后用10个预测集样本对模型的性能进行验证,决定系数为0.9495,预测集标准偏差为0.0547,相对分析误差为4.45,表明该方法能应用于生产实践。最后选取样品的近红外光谱的6502.8—3999.6cm-1波段建立9个厂家的益母草颗粒聚类分析模型,结果不同厂家的样品能很好地进行分类,然后使用预测集样品对该模型进行检验,结果表明该方法准确快速。
The quantitative analysis model was established for the determination of Leonurus alkaloids by near infrared spectroscopy and partial least squares. The model had a coefficient of determination of 0.9622 and the mean square error of internal cross validation was 0.0576. Then, The performance was verified with a coefficient of determination of 0.9495, a standard deviation of prediction set of 0.0547 and a relative analysis error of 4.45, indicating that the method can be applied to production practice. At last, the model of the motherwort particle aggregation of nine factories was established by the 6502.8-3999.6cm-1 band in the near infrared spectrum of the sample. As a result, the samples from different manufacturers can be well classified, and then the model was tested by using the prediction set sample. The results show that the method is accurate and rapid.