基于近红外光谱技术的云南片烟综合质量评价

为探索利用近红外光谱技术快捷、全面评价片烟质量的可行性，以2015—2017年初烤烟叶和复烤片烟为研究材料，使用偏最小二乘法等数据处理方法建立了片烟常规化学成分、香型、部位、感官质量的近红外光谱预测模型，验证模型效果后对2018—2019年度云南片烟质量进行了预测和评价。结果表明，2019年度较2018年度，化学成分呈现总糖、还原糖极显著下降、总氮极显著上升的趋势；清香型指数略有下降；两年上、中、下部位指标差异不显著；近红外感官质量模型评价的70个片烟中有57个片烟与感官评价结果相符，预测准确率达到81.4%，且年度间差异极显著；近红外预测结果与感官评吸结果具有较高的一致性。利用近红外技术评价片烟综合质量可以为卷烟配方设计提供参考。

Comprehensive Quality Evaluation of YunnanTobacco Strips Based on Near Infrared Spectroscopy

In order to explore the feasibility of using near-infrared spectroscopy to evaluate the quality of tobacco strips quickly and comprehensively, flue-cured tobacco and tobacco strips from 2015 to 2017 were used as research materials. Prediction models of routine chemical compositions, aroma type, position and sensory quality were established by the partial least square method. The quality of Yunnan tobacco strips in 2018-2019 was used for comparative evaluation after verifying the effect of the model. The results showed that compared with 2018, the total sugar and reducing sugar decreased significantly in 2019, while total nitrogen increased significantly; The fresh aroma index decreased slightly. There was no significant difference in the index of upper, middle and lower positions between the two years. 57 of the 70 tobacco strips evaluated by the near-infrared spectroscopy sensory quality model were consistent with sensory evaluation, with a prediction accuracy of 81.4% and highly significant differences between years. The near-infrared spectroscopy prediction results were in high agreement with the sensory evaluation results. The use of near-infrared spectroscopy to evaluate the comprehensive quality of tobacco strips can provide a reference for cigarette formulation design.