Industrial Data-Driven Model of Tobacco Leaf Loose Moisture Regain

Abstract

With the development of advanced information technology such as cloud computing, big data, micro services and artificial intelligence, the manufacturing industry has also tried to introduce a large number of these technologies to achieve industrial transformation and upgrading. In the process of cigarette processing, there are many pain points that cannot be solved by the traditional manufacturing process, such as the loose moisture control in the silk making process and the balance control of the temperature, hot air and moisture in the thin sheet drying cylinder. The traditional manufacturing process requires continuous pre drying control based on experience. Due to the differences in the process level of manual operation, the quality of cut tobacco produced is also uneven, which can't meet the requirements of homogenization of cigarette production. Therefore, this paper proposes a loose regain model of tobacco leaf based on industrial data. By analyzing the dependence of various processing parameters, we obtain important parameters that affect the amount of water added to the loose regain, use these parameters to learn the characteristics of the loose regain process, automatically predict the amount of water added, and improve the cigarette homogenization processing level. Experiments show that our proposed model can effectively improve the quality rate of products. Finally, based on the CPS architecture and the proposed model, an intelligent control system of loose moisture regain is developed, which has been successfully deployed in the actual production process..