Analysis And Research On Multi-source Data Modeling Of Threshing Process

In recent years,with the goal of high-quality development proposed by the tobacco industry,there are higher requirements for the refinement,digitization,and modernization of cigarette processing processes.The threshing section is one of the key links in the threshing and redrying process.The quality of the tobacco leaf shape after threshing is related to the processing effects of the wind separation section,the redrying section,and the cigarette shredding process.Therefore,a multi-source data model is established based on the equipment data,sheet structure characteristics testing data,and raw material physical characteristics data of tobacco leaves in the threshing section.The relationship between process data and threshing quality during the production process is analyzed,and the distribution ratio of sheet structure characteristics after threshing under certain conditions is predicted.The results provides a solution for adjusting process parameters,improving the threshing effect,and promoting cost reduction and efficiency increase in the threshing and redrying workshop.Based on the production data of the first tobacco threshing and redrying workshop in a cigarette factory,this paper selects various types of data from three representative characteristic grades of tobacco raw materials in the threshing section as the analysis object,and studies the process parameters,such as the moisture content of the second moisture discharge,the temperature of the second moisture discharge,the frequency of the first beater,the frequency of the second beater,the frequency of the third beater,the frequency of the first air distributor,and the frequency of the second air distributor The influence of the frequency of the ninth wind turbine on the sheet shape structure indicators(large sheet rate,medium sheet rate,small sheet rate,fragment rate,and leaf stem content rate).Using the principal component analysis method to select key parameters from many parameters,and using the key parameters as input variables,using Python and Matlab software to establish three prediction models including random forest,support vector machine,and BP neural network,to predict the quality of tobacco leaf after threshing,calculate the mean square error and root mean square error of the models,and analyze and compare the prediction effects of different models.Based on the comparison results,the BP neural network model is superior to the other two machine learning models.According to multiple linear regression and multi-objective optimization algorithms,an optimization scheme for the process parameters of the threshing process was obtained,and validation experiments were conducted on the main line of the threshing section in a certain redrying workshop.The results showed the feasibility of the optimization plan,providing a basis for subsequent practical applications.In this paper,the internal relationship between the key process parameters in the threshing section and the shape quality of the threshing tobacco leaves was established,a suitable prediction model for the shape index was determined,and multiple sets of process parameter optimization schemes were designed and verified by experiments,which has significant reference value and useful purpose for adjusting the parameters of threshing process equipment and improving the quality of threshed leaves in actual production.