基于近红外光谱及组合间隔偏最小二乘法的天南星中水分及总黄酮含量测定研究

摘要

目的 建立天南星中水分及总黄酮的近红外含量预测模型.方法 采用近红外漫反射光谱,利用TQ Analyst 9.0软件进行光谱预处理,并结合Matlab平台下的组合间隔偏最小二乘法进行波段的优选.以校正集相关系数(Rc2)、校正集预测均方根误差(RMSEC)、验证集相关系数(Rp2)、验证集预测均方根误差(RMSEP)为指标评价模型的优劣.结果 水分较为理想的建模波段范围为4613.6~4968.4 cm-1、8871.2~9226 cm-1、10299.4~10645.2 cm-1;总黄酮建模波段范围为4376.3~4849.6 cm-1、5796.2~6269.5 cm-1、8182.6~8635.9 cm-1.建模后水分的含量预测模型Rc2=0.9562,RMSEC=0.332,Rp2=0.9052,RMSEP=0.554;总黄酮含量预测模型为Rc2=0.9377,RMSEC=0.021,Rp2=0.8988,RMSEP=0.027.结论 建立的近红外光谱测定方法 操作简便、快速、准确,为水分及总黄酮含量的快速检测提供了借鉴.%Objective To establish the prediction model for near infrared spectroscopy (NIR) content determination of moisture and total flavonoids of Arisaematis Rhizoma.Methods NIR was applied on TQ Analyst 9.0 software and together with synergy interval partial least squares (SIPLS) program in Matlab platform to optimize waveband. The evaluation index of prediction model included root mean square error of calibration (RMSEC), correlation coefficient of calibration (Rc2), root mean square error of prediction (RMSEP) and the correlation coefficient of prediction (Rp2).Results The characteristic wavelength regions for measurement of moisture were 4613.6–4968.4 cm-1, 8871.2–9226 cm-1, and 10299.4–10645.2 cm-1. While for total flavonoids, the sensitive wavelength regions were 4376.3–4849.6 cm-1, 5796.2–6269.5 cm-1, and 8182.6–8635.9 cm-1. As a result, the Rc2, RMSEC, Rp2 and RMSEP values for prediction of moisture were 0.9562, 0.332, 0.9052 and 0.554, respectively. And the Rc2, RMSEC, Rp2 and RMSEP values for prediction of total flavoniods were 0.9377, 0.021, 0.8988 and 0.027, respectively.ConclusionThe established method is simple, rapid and accurate, which can be used as an option for rapid quality evaluation of moisture and total flavonoids of Arisaematis Rhizoma.