为了提高检测效率,采用近红外光谱结合偏最小二乘法(PLS)建立了快速预测胆宁片中间体中大黄素含量的定量分析模型。使用近红外光谱仪采集136批训练集样品的原始近红外光谱图,以标准正态变量(SNV)+二阶导数(2^nd)对原始近红外光谱数据进行预处理,在1 300~1 764 nm和1 932~2 200 nm两个波长范围内以PLS建立大黄素的定量分析模型。结果显示,模型的决定系数(R²)为0.770 8,交叉验证均方根误差(RMSECV)为0.086 2。用199批测试集样品进行验证,结果显示,模型分析得到的预测值与高效液相色谱法分析得到的参考值的相对平均偏差绝对值均不大于5.0%,表明该模型预测准确度较高。

In order to improve the detection efficiency, a quantitative analysis model for rapid prediction of emodin in Danning tablet intermediate was established by near infrared spectroscopy with partial least square method (PLS). The original near infrared spectra of 136 batches of training set samples were collected by near infrared spectrometer, and the original near infrared spectra data were pretreated by standard normal variate (SNV) and second derivative (2^nd), and the quantitative analysis model of emodin was established by PLS in the wavelength ranges of 1 300-1 764 nm and 1 932-2 200 nm. It was shown that the determination coefficient (R²) of the model was 0.770 8 and the root mean square error of cross validation (RMSECV) was 0.086 2. The 199 batches of testing set samples were used for verification. The results showed that the relative mean deviation absolute values of the predicted values obtained from the model analysis and the reference values obtained from the high performance liquid chromatography analysis were less than 5.0%, indicating that the prediction accuracy of the model was high.