Study on the correlation between softness and physical quality of formula tobacco in cigarettes of different specifications
[Objective]This study aims to explore the relationship between the softness and physical quality of formula tobacco in cigarettes of different specifications.[Methods]Variance analysis,correlation analysis,and grey relational analysis were conducted on the softness and physical quality of formula tobacco in regular,king-size,and slim cigarettes,establishing an XGBoost prediction model for the softness of formula tobacco in cigarettes of different specifications.[Results](1)For cigarettes of the same specification,as the price increases,the softness value decreases,and the level of softness improves;at the same price level,the order of softness level of formula tobacco in cigarettes of different specifications is:slim>king-size>regular;there is a significant difference in the softness of formula tobacco among different specifications.(2)The softness values of formula tobacco in cigarettes of different specifications are highly negatively correlated with the rates of long,medium,and whole-stalk tobacco,and elasticity,and highly positively correlated with filling value,moisture content,and cut tobacco rate.(3)The grey relational degree of the physical quality of formula tobacco to its softness in cigarettes of different specifications is in the order of:first class(medium stalk rate,moisture content)>second class(filling value and cut tobacco rate)>third class(long stalk rate,whole-stalk rate,elasticity).(4)The XGBoost models established for regular,king-size,and slim cigarettes'formula tobacco predict with accuracies of 96.2%,92.4%,and 95.4%respectively,indicating that the XGBoost prediction models are stable and reliable.[Conclusion]The dynamic control of tobacco softness can be achieved to improve tobacco quality,and can also provide technical support for improving the applicability of tobacco.
cigarettes of different specificationsformula tobaccosoftnessphysical qualityXGBoost model
国家科技期刊平台
NSTL
万方数据
