Research on the relationship between rebound characteristics of cut tobacco and physical indicators of cigarettes based on machine learning
[Background]This study aims to clarify the impact of the rebound characteristics of cut tobacco on the physical quality of cigarettes.[Methods]The rebound characteristics of cut tobacco of different brands and batches and the physical quality of cigarettes from corresponding production batches(cigarette hardness,cigarette dust content,and end loose filler)were analyzed for correlation.Based on the physical indicators of cut tobacco(long shred rate,medium shred rate,short shred rate,filling power,moisture content)and rebound characteristics,machine learning was used to build and optimize prediction models for the physical indicators of cigarettes(cigarette hardness,cigarette dust content,and end loose filler).[Results](1)The rebound characteristics of cut tobacco were significantly positively correlated with cigarette hardness and negatively correlated with cigarette dust content and end loose filler.(2)The best prediction model for cigarette hardness was gradient boosting regression optimized by grid search,with an R2 of 0.95;the best prediction model for cigarette dust content was gradient boosting regression optimized by Bayesian optimization,with an R2 of 0.97;the best prediction model for end loose filler was random forest regression optimized by grid search,with an R2 of 0.97.[Conclusion]The constructed models all have high accuracy and can be used to predict the physical indicators of cigarettes under certain experimental conditions.
rebound characteristicsphysical indicators of cut tobaccophysical indicators of cigarettesmachine learning
万方数据
维普
