Research on the influence of physical indicators of cut tobacco on its rebound characteristics.

Using 5 grades of conventional cigarette finished tobacco as the research object, multiple linear regression analysis method was used to study the influence of physical indicators of cut tobacco on its rebound characteristics, and a prediction model for tobacco rebound characteristics based on BP neural network was constructed. The results showed that the degree of influence of physical indicators of cut tobacco on the rebound characteristics of tobacco was ranked from large to small as follows: broken tobacco rate>medium tobacco rate>long tobacco rate>filling value>elasticity>moisture content. Among them, medium tobacco rate, long tobacco rate, elasticity, and moisture content were positively correlated with the rebound characteristics of tobacco, while broken tobacco rate and filling value were negatively correlated with the rebound characteristics of tobacco. In the constructed BP neural network prediction model, the comparison between the predicted value of the test set and the true value R² was 0.965 7, the overall model accuracy reached 98. 10%. The difference between the predicted and measured values of cut tobacco rebound characteristics was small, and the model had high prediction accuracy and reliability, which could be used for accurate estimation of cut tobacco rebound characteristics. [ABSTRACT FROM AUTHOR]