基于叶面特征的烤烟中部叶采收成熟度量化初探  

A preliminary investigation on the quantification of the harvesting maturity of flue-cured tobacco middle leaf based on leaf surface characteristics

作　　者：裴晓东 赵晨 武劲草 张文军 李帆 王建文 刘优雄 李姣姣 李生栋 路晓崇[2] PEI Xiaodong;ZHAO Chen;WU Jincao;ZHANG Wenjun;LI Fan;WANG Jianwen;LIU Youxiong;LI Jiaojiao;LI Shengdong;LU Xiaochong(Changsha City Corporation of China National Tobacco Corporation,Changsha Hunan 410000,China;College of Tobacco,Henan Agricultural University,Zhengzhou,Henan 450002,China;Jiangsu China Tobacco Industry Company Limited,Nanjing,Jiangsu 210019,China;Qujing City Corporation of Yunnan Province Tobacco Company,Qujing,Yunnan 655000,China)

机构地区：[1]中国烟草总公司长沙市公司,湖南长沙410000 [2]河南农业大学烟草学院,河南郑州450002 [3]江苏中烟工业有限责任公司,江苏南京210019 [4]云南省烟草公司曲靖市公司,云南曲靖655000

出　　处：《中国烟草学报》2024年第1期46-54,共9页Acta Tabacaria Sinica

基　　金：中国烟草总公司长沙市公司资助项目“烟叶密集烘烤增容提质关键技术研究”(2022430100240141)。

摘　　要：【目的】推进烤烟采收成熟度数字化判断的实现。【方法】以云烟87中部烟叶为试验材料,将采集的746个烟叶样品按照欠熟、尚熟、适熟和过熟4个档次进行分类,利用模糊DEMATEL分析法对烤烟的农艺特征、颜色特征以及纹理特征3方面的叶面特征指标进行权重赋予,初步得到不同成熟度烤烟的得分区域,之后采用偏最小二乘判别(PLS-DA)法简化烤烟采收成熟度量化模型,并对该模型的鲁棒性进行验证。【结果】(1)欠熟档次的最终得分确定区域为[0.0,81.0)分,尚熟档次确定区域为[81.0,85.0)分,成熟档次确定区域为[85.0,91.0]分,过熟档次确定区域为(91.0,100]分;(2)所构建的成熟度识别模型的正确率达到89.04%。【结论】通过采集鲜烟叶的叶面特征可以对烤烟的进行成熟度量化。[Objective]This study aims to promote the realization of digital judgment of flue-cured tobacco harvesting maturity.[Methods]746 tobacco samples collected from the central part of Yunnan Tobacco 87 were classified according to four grades:under-ripe,still-ripe,moderately ripe and over-ripe,and the fuzzy DEMATEL analysis method was used to assign weights to the leaf surface characteristics of flue-cured tobacco in three aspects:agronomic characteristics,color characteristics and texture characteristics.After that,the partial least squares discriminant(PLS-DA)method was used to simplify the quantification model of the harvesting maturity of the flue-cured tobacco,and the robustness of the model was verified.[Results]The results showed that:(1)the final score for the under-ripe grade was determined as[0.0,81.0),and the final score for the under-ripe grade was determined as[0.0,81.0).(2)The final scores of under-ripe grades were determined as[0.0,81.0),under-ripe grades as[81.0,85.0),under-ripe grades as[85.0,91.0],and over-ripe grades as(91.0,100];(3)The results of the optimized maturity recognition model had a correct rate of 89.04%.[Conclusion]The maturity of flue-cured tobacco was quantified by collecting the leaf surface characteristics of fresh tobacco leaves.

关 键 词：烤烟 成熟度量化 模糊DEMATEL分析 偏最小二乘判别 

分 类 号：TS452[农业科学—烟草工业]