Research On Digital Twin Modeling And Process Quality Prediction Of Tobacco Drying Process

The drying process of tobacco is one of the core processes in the process of making tobacco.The quality,taste and smoke of tobacco will be affected by the processing level.The tobacco drying process is complex and changeable,and there are a large number of drying process parameters,which are coupled and influenced by each other.In addition,the material flow,energy flow and information flow in the tobacco drying process are still isolated from each other in the processing stage,and the data of information space and physical space lack interaction and integration,so the level of interaction,autonomy and predictability in the production and processing process is still low.Thus,it seriously restricts the further improvement of the intelligent level of tobacco drying process.At present,the intelligent improvement of tobacco drying process mainly focuses on production process data monitoring and process quality prediction based on statistical analysis and mathematical model.However,the traditional production process data monitoring has some problems,such as low transparency and visibility,poor real-time interaction,and single monitoring method.The process quality prediction based on statistical analysis and mathematical model only predicts a single quality index of export moisture content,which makes the intelligence,prediction,transparency and real-time interaction of tobacco drying process insufficient,and it is difficult to meet the intelligent upgrading and transformation of tobacco drying process.In this thesis,a digital twin model of tobacco drying process based on the concept of digital twin technology was established,aiming at the problems of low visualization degree,opaque operation process,information interaction and delayed process quality prediction of tobacco drying process.The visual monitoring and quality prediction of tobacco drying process were studied.Firstly,by analyzing the tobacco drying process in detail,based on the concept of digital twin model construction,from three dimensions of geometry,logic and data,the digital twin model of tobacco drying process was established,which included visual geometry model,interactive logic model and computational data model,and realized the multi-dimensional description of tobacco drying process.Secondly,aiming at the problem of information interaction in tobacco drying process,a data-driven virtual-real interaction technology in tobacco drying process was explored,and a virtual-real interaction technology system including data acquisition,data transmission,virtual-real mapping and feedback control was established.Based on the establishment of virtual and real interaction channel,the characteristics of process parameters of tobacco drying process were deeply analyzed,and the prediction model of process quality was constructed by deep learning neural network,which provided the model basis for the realization of online prediction of process quality.Finally,based on the above model basis and technology basis,the digital twin system of tobacco drying process was developed,and the 3D visualization real-time monitoring of tobacco drying process and online prediction of process quality were realized.The effectiveness and practicability of the proposed method were verified by the tobacco drying process of a tobacco factory.The research results of this thesis provide a new technical means for improving the intelligent level of tobacco drying process,and have important application value for improving the production efficiency and quality stability of tobacco drying process,and lay a technical foundation for realizing the parameter optimization and feedback control of tobacco drying process.