论文部分内容阅读
目的应用近红外光谱法和数据分析软件,对金银花中水分含量进行快速测定。方法利用甲苯法测定样品中水分的含量,运用偏最小二乘(PLS)法建立其含量与NIR光谱之间的多元校正模型,对未知样品进行含量预测。结果建立的水分校正模型相关系数(R2)、估计误差均方根(RMSEE)、相对分析误差(RPD)分别为0.933,0.18%,3.86。经外部验证,校正模型的预测均方差(RMSEP)、平均回收率分别为0.216,98.9%。结论此方法具有快速简便、准确无损的特点,可应用于金银花中水分含量的快速检测。
Objective To determine the moisture content of honeysuckle by near-infrared spectroscopy and data analysis software. Methods The content of water in the samples was determined by toluene method. The multivariate calibration model between the content and NIR spectra was established by partial least squares (PLS) method to predict the content of unknown samples. Results Correlation coefficient (R2), root mean square error of estimation (RMSEE) and relative analysis error (RPD) of water calibration model were 0.933,0.18% and 3.86 respectively. After external validation, the mean square error (RMSEP) of the calibration model and the average recovery rates were 0.216 and 98.9%, respectively. Conclusion This method is rapid, simple, accurate and non-destructive, and can be applied to the rapid determination of water content in honeysuckle.