近红外光谱结合导数方法在烟叶等级一致性检测中的应用

摘要

It was very important to improve tobacco leaf quality detection method in tobacco industry.The traditional detection methods based on appearance and chemical were time-consuming and laborious,and it is difficult to meet the requirements of industrial production.A total of 78 batches of tobacco leaves of Guiyang rebaked plant on 2016 were used in this study.Near Infrared spectroscopy,nicotine,total sugar and other chemical data were analyzed.Based on the modern spectral signal processing techniques such as derivative and the coefficient of variation,a fast detection method based on near infrared spectroscopy was developed.The results showed that using the full spectral average at calculating variation and the first derivative spectra,the correlation coefficient between spectral variation coefficient and the nicotine vaiue's variation coefficient reached 86.60 ％ and 86.12 ％,respectively.This new method could be used in tobacco grade detection directly.%针对目前实际生产应用中,烟叶等级一致性检测流程过度依赖传统的外观评价方法和化学检测方法而导致的成本过高,效率低下等问题,采用导数等现代光谱信号处理技术和变异系数等统计指标,对2016年度贵阳省内共计78组烟叶样品的近红外光谱和烟碱、总糖等化学值数据进行分析与研究.结果表明:采用全谱段均值进行变异系数计算,以及一阶导数处理后,光谱变异系数与尼古丁值变异系数的相关系数达到86.60％,与总糖值变异系数的相关系数达到86.12％,满足实际生产应用中对烟叶等级一致性检测的准确度要求,能够实现直接应用近红外光谱进行烟叶等级一致性检测,同时降低烟叶等级一致性检测的人力物力成本,达到绿色快速无损检测的目的.