带梗烟叶在线近红外检测模型的建立与应用研究

摘要

The model date was collected from calibrated spectrum in the form of fiber-optic diffuse reflection and chemical reference values, using online near-infrared spectrum testing and the usage of PLS, established an online near-infrared spectrum detection model for rapid detection on the nicotine and total sugar of production line in tobacco leaves with stems.The results showed that:compared with external, the average absolute deviation of nicotine and total sugar were 0.17%and 1.25%while the relative deviation were 7.14%and 4.14%, respectively.The coefficient of variation ( CV) of the nicotine values in tobacco prod-uct could be control to below 3%by using online near-infrared spectrum data to guide recipe leaf beating.By building up the model of online near-infrared spectrum for threshing and redrying in tobacco, which could control and guide the feeding of raw material of tobacco effectively, and improve homogeneity of the chemical composition of the finished smoke.%通过在线采集校正样品，利用偏最小二乘法（ PLS）分别建立了带梗烟叶中烟碱和总糖的在线近红外光谱检测模型。与外部比对发现，烟碱和总糖绝对偏差的平均值分别为0．17％和1．25％；相对偏差的平均值分别为7．14％和4．14％；利用在线检测的数据指导配方打叶，能将成品片烟烟碱的变异系数控制在3％以下。表明建立打叶复烤烟叶分选环节的在线近红外检测模型，能够有效地调控和指导原烟的投料，提升成品片烟化学成分的均匀性。