摘要
用近红外(NIR)光谱技术直接扫描烟叶粉末,通过一阶导数+附加散射校正和偏最小二乘法(PLS)的处理,建立了预测烟草pH值的数学模型。该模型的内部验证决定系数(r^2)为0.8333,预测标准差(RMSECV)为0.0603;外部验证决定系数(r^2)为0.804,预测标准差(RMSEP)为0.0926;PLS模型的因子数为7。通过配对-t检验,该方法与标准方法测定的结果无显著性差异。该方法精密度高,可用于烟草pH值的快速预测,为烟草行业节省大量资金和显著提高工作效率。
A mathematical model of pH of tobacco was established by scanning tobacco powder with NIR spectrometry. The determination were pretreated by first derivative ＋ add-scatter Correction and partial least square （PLS）. The determination coefficient （r2） was 0. 8333 in cross validation and the root mean square error of cross validation （RMSECV） was 0. 0603, the determination coefficient （r2） was 0. 804 in test set validation and the values of root mean square error of prediction（RMSEP） for pH of tobacco was 0. 0926,the factor number for PLS model was 7. The results of prediction were compared with those of standard methods with no significant difference. NIR analysis has advantages of rapidity and precision,and can be applied for rapid determination of pH of tobacco. The method would save great costs and enhance efficiency for tobacco and cigarette industry.
出处
《光谱实验室》
CAS
CSCD
北大核心
2010年第6期2266-2270,共5页
Chinese Journal of Spectroscopy Laboratory
关键词
近红外光谱
烟草
PH
Near Infrared Spectroscopy
Tobacco
pH
