基于特征分布和判别分析的卷烟梗签识别方法建立与评价  

Establishment and Evaluation of a Cigarette Stick Identification Method Based on Characteristic Value Distribution and Discriminant Analysis

作　　者：王瑞雪 朱强强[1] 唐习书 刘泽 黄业伟[1] 祁林[2] WANG Ruixue;ZHU Qiangqiang;TANG Xishu;LIU Ze;HUANG Yewei;QI Lin(Yunnan Agricultural University,Kunming,Yunnan 650201,China;Hongyun Honghe Tobacco(Group)Co.,Ltd.,Kunming,Yunnan 650231,China;Technology Center,China Tobacco Yunnan Industrial Co.,Ltd.,Kunming,Yunnan 650231,China)

机构地区：[1]云南农业大学,云南昆明650201 [2]红云红河烟草(集团)有限责任公司,云南昆明650231 [3]云南中烟工业有限责任公司技术中心,云南昆明650231

出　　处：《热带农业科学》2024年第11期104-110,共7页Chinese Journal of Tropical Agriculture

基　　金：红云红河烟草(集团)有限责任公司科技计划项目(No.HYHH2020GY03)。

摘　　要：识别出卷烟各生产环节中梗签分布特征,并建立模型识别出特定环节的梗签,为后续有效剔除梗签、降低梗签含量提供依据。以常规卷烟(A牌号)和中支卷烟(B牌号)为研究对象,分别对制丝、卷制、跑条环节中的梗签进行抽检,测定梗签宽度、长度、面积、颜色L值,分析各环节中梗签特征的分布频率,通过提取特征值与均匀性系数建立判别分析模型。结果显示,A、B牌号跑条梗签在宽度、长度分布比例最大的区间分别为5~10、4~6和0.8~1.2、1.2~1.4 mm,2种卷烟的跑条梗签面积和L值分布比例最大区间一致,分别为5~10 mm^(2)、10~12;提取特征值和均匀性系数发现,2种卷烟跑条梗签区别于其他梗签的显著指标是梗签面积,大小顺序为跑条<卷制<制丝。利用不同类型卷烟各生产环节梗签分布特征值和均匀性系数建立的判别函数模型,可实现原始分组梗签识别正确率高于90%,交叉个案识别正确率高于85%,达到工业生产要求。To identify the distribution characteristics of sticks in each production process of cigarettes and establish a model to identify sticks in specific processes to provide a basis for effectively removing sticks and reducing the stick content.With conventional cigarettes(brand A)and medium cigarettes(brand B)as the research objects,the sticks used in the process of slicing,rolling,and stripping were sampled to determine the width,length,area,and color L value of the sticks;analyze the distribution frequency of stick characteristics in each process;and establish a discriminant analysis model by extracting the characteristic values and uniformity coefficient.The results revealed that the sections with the largest width and length distri-bution ratios of brands A and B were 5–10,4–6,0.8–1.2,and 1.2–1.4 mm,respectively.The area and L value distribution ratio of the two kinds of cigarettes were the same:5–10 mm^(2) and 10–12,respectively.The extraction of characteristic values and uniformity coefficients revealed that the significant indexes of the stick area and size sequence were different from those of the other processes in the following order:stripping<rolling<slicing.The discriminant function model,which is based on the characteristic value and uniformity coefficient of the stick distribution in different production processes,achieves a more than 90% correct rate of stick identification for the original group,and the cross-case identification is greater than 85%,which can meet the requirements of industrial production.

关 键 词：梗签识别 特征值 均匀性系数 判别分析 卷烟 

分 类 号：S572[农业科学—烟草工业]