Predicting the Reconstituted Stem Moisture by Improved Decision Tree

Moisture is an important control target in the tobacco industry. The moisture of reconstituted stems has an important influence on the sensory and process quality of cigarette. In the process of reconstituted stalks production, in order to ensure the moisture standard, the operators mainly rely on a large amount of experience for controlling, and different operators have different deviation in real-time adjustment under disparate formula. Meanwhile, due to the low level of the automation technology and the slow respond of the cigarette machine, it is easy to cause large moisture fluctuations and directly lead to unqualified products. For the purpose of solving the problem, this paper proposes a moisture prediction model based on Apriori algorithm and decision tree. This model inputs historical data samples to achieve the moisture prediction of each important process section, especially the width of 0.11mm stem, the deviation of the moisture prediction is only 0.326 comparing to the standard. After test and verifying, this model can initially replace human labor and realize intelligent automatic control.