Establishment and Evaluation of a Cigarette Stick Identification Method Based on Characteristic Value Distribution and Discriminant Analysis
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 mm2 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.
万方数据
维普
