摘要
应用近红外光谱技术建立烟草17项主要化学成分的快速无损检测方法。收集700个具有代表性样品的光谱,建立其相应指标的近红外模型。在所有的校正模型中,原始谱图经过一阶导数和偏最小二乘(PLS)处理,大约50个外部样品用于所建模型的验证。烟草中总挥发酸、总挥发碱、石油醚提取物总量、石油醚提取物中性成分、多酚、淀粉、纤维素、硫酸根、pH、灰分、水溶性灰分碱度、总糖、还原糖、总氮、生物碱、氯、钾等十七项指标的预测标准偏差(RMSEP)分别为0.020、0.009、0.402、0.393、0.578、0.583、0.932、0.139、0.117、0.634、0.235、1.720、1.407、0.104、0.173、0.037和0.300。该结果表明近红外光谱技术在分析17项烟草化学指标时均可以替代经典化学方法。作为一种质量控制方法,近红外光谱技术的应用将为烟草行业节省大量的资金和显著提高工作效率。
FT-NIR was evaluated as a method to simultaneously analyze 17 chemical components in tobacco leaf. Spectra of 700 typical samples were collected to develop the calibration models, among which the raw spectra were pretreated with first derivative and partial least square. About 50 randomly selected validation samples were used to test the accuracy of each calibration model. The values of root mean square error of prediction （RMSEP） for total volatile acids, total volatile bases, total petroleum ether extracts, petroleum ether extracts neutral, polyphenol, starch, cellulose, sulfate, pH, ash, water soluble ash bases, total sugar, reductive sugar, total nitrogen, alkaloids, chlorine, and potassium were 0.020, 0.009, 0.402, 0.393, 0.578, 0.583, 0.932, 0. 139, 0. 117, 0.634, 0.235, 1.720, 1.407, 0.104, 0. 173, 0.037, 0.300 respectively. The satisfactory results indicated that FT-NIR could be an alternative to the traditional wet chemical method to simultaneously analyze the above 17 chemical components in tobacco leaf. The utilization of FT-NIR as a quality control method will reduce costs and enhance efficiency for the tobacco industry.
出处
《中国烟草学报》
EI
CAS
CSCD
2006年第2期8-12,共5页
Acta Tabacaria Sinica
基金
国家烟草专卖局资助项目(国烟科[2003]542号)的一部分
关键词
近红外
烟草
快速分析
near infrared
tobacco
simultaneous analysis
作者简介
蒋锦锋（1979-）,男,郑州烟草研究院在读研究生.主要从事烟草化学的常规检测及近红外光谱技术分析研究.E-mail：jinfengzzu@yahoo.com.cn.郑州枫杨街2号
