本平台为互联网非涉密平台,严禁处理、传输国家秘密或工作秘密

云南不同产地烟叶物理特性差异分析

Research on the differences in physical characteristics of flue-cured tobacco leaves in different producing areas from Yunnan

  • 摘要: 为探究云南不同产地烟叶物理特性差异,采集云南13个植烟州市3 646份上、中、下部烤烟烟叶样本,测定其拉力、厚度、含梗率,并计算拉厚比。采用多重比较、聚类分析和偏最小二乘判别分析(PLS-DA)等手段,对比云南不同产地烟叶的异同及共性,初步建立云南烟叶小产地识别模型。结果表明:(1)烟叶拉力与厚度呈正相关关系,含梗率与厚度在中、下部烟叶中呈负相关关系。(2)不同产地烟叶物理指标差异显著,上部烟叶中以昭通烟叶拉力与厚度最高,中部烟叶以丽江烟叶拉力最高,而下部烟叶以德宏烟叶拉力与厚度最高;德宏烟叶含梗率最低;大理、丽江烟叶拉厚比突出。(3)基于物理指标并结合自主取样策略,构建了云南省4个产地烟叶的判别模型。随机森林与PLS-DA模型在测试集上的最高准确率分别达到93.50%和87.75%,且两种模型均以拉厚比为最重要的判别变量。(4)聚类分析将云南烟叶分为五类,大理与丽江烟叶始终同类,呈现“高拉厚比”特征;德宏烟叶始终独立为类,含梗率最低。

     

    Abstract: This study aimed to investigate the differences of physical characteristics of flue-cured tobacco leaves (FTLs) across different production areas from Yunnan province. A total of 3,646 FTLs samples from 13 tobacco-planting cities/autonomous prefectures in Yunnan were collected, including upper, middle and lower parts. Tensile force, thickness, and stem content were measured, and then the tensile force-to-thickness ratio was calculated. Multiple comparisons, cluster analysis and partial least squares discriminant analysis (PLS-DA) were employed to reveal the similarities and differences of FTLs in different production areas from Yunnan, and a preliminary identification model was established for specific regions within the province. The results showed that: (1) Tensile force showed positively correlation with thickness in all parts of FTLs, while stem content exhibited negatively correlation with thickness in middle and lower leaves. (2) Physical indices differed significantly among production areas: FTLs from Zhaotong showed the highest tensile force and thickness in upper leaves, that from Lijiang had the highest tensile force in middle leaves. While FTLs from Dehong exhibited the highest tensile force and thickness in lower leaves. Regarding stem content, FTLs from Dehong showed the lowest value in all parts. FTLs from Dali and Lijiang displayed notably higher tensile force-to-thickness ratio. (3) Based on physical indices combined with bootstrap resampling approach, discrimination models including Random Forest and PLS-DA were constructed for FTLs from four production regions in Yunnan Province, achieving accuracies of 93.50% and 87.75%, respectively, on the test set. In both models, the tensile-to-thickness ratio was identified as the most important discriminant variable. (4) Cluster analysis revealed that FTLs from Yunnan could be classified into five categories. FTLs from Dali and Lijiang were always of the same category, exhibiting the characteristics of "high tensile force-to-thickness ratio". That from Dehong have always been classified as an independent category, with the lowest stem content.

     

/

返回文章
返回