Physical Quality of Tobacco Production and Primary Processing Neural Network Model and Its Evaluation

In order to study the effect of primary processing intensity on the physical quality of tobacco, for the productive process of a certain brand of cigarettes, the process includes loosening and conditioning, the first feeding, the increasing temperature and humidity on the leaves, and drying process control against moisture, five important process parameters were used for uniform design. The filling value of tobacco, whole tobacco ratio, kibbling tobacco ratio of each group was detected. Then,the BP neural network model of the process parameters and quality indexes were set. The result showed that whole tobacco ratios were 80.34%, 82.73%, 85.42%, the corresponding predicted results were 80.89%, 82.12%, 84.86%; kibbling tobacco ratios were 1.87%, 1.61%,1.74%, the corresponding predicted results were 1.6%, 1.58%, 2.2%; filling value was 4.57,4.73,4.60 cm3/g, the corresponding predicted results were 4.48,4.51,4.62 cm3/g. It supplies new methods for optimization of tobacco process parameters, can be used in the production of guidance, reduce the cost of test research.