The Research On Production Quality Forecast Of Cigarette Making Machine Based On Important Operating Condition Parameters

In recent years,the expedited integration between the new information technologies such as the Internet,big data and artificial intelligence and the real economy drive the manufacture industry to further evolve towards digitalization,networking and Intellectualization,which provides the important opportunities and challenges for the structural adjustment and industrial upgrade of the manufacturing industry.As an important pillar industry of the national economy,the tobacco industry should take the initiative to integrate into the theme of high-quality development of the country.We should not only strengthen responsibility,make a difference and make full use of cigarette machine to produce big data,but also extract valuable information to make the production more intelligent.Taking the practical big data of a brand cigarette produced by protos1-8 cigarette machine as an example,this paper makes some exploratory analysis on the evaluation and prediction of the production process by using the statistical analysis method of big data.Firstly,I select a series of indicators including physical measurement,gas source appearance detection and waste removal rate in the cigarette production process,and use the method of The Grey Correlation Analysis and Fuzzy Comprehensive Evaluation to comprehensively evaluate the cigarette quality,to compare the pros and cons of these two comprehensive evaluation methods which makes preparation for the selection of important working condition variables and prediction.Secondly,more than 50 million single cigarette records in the production process were used to analyze the basic characteristics of each working condition parameter through descriptive statistics.the empirical analysis based on Lasso method,BP neural network based variable selection methods is used to screen the working condition of important variables which affect the quality of cigarette-rolling machines production and get the working condition of important variables affecting the quality of cigarette-rolling machines production.The order of these variables are: VE big fan pressure and suction ribbons actual location,ribbon position standard deviation,the SRM compaction end position,M12 V motor speed,VE a winnowing pressure temperature,cut tobacco and tobacco moisture.Finally,use the important working condition parameters to predict the production quality of cigarette rolling machine,optimize the model parameters with The Bayesian Optimization Algorithm,and compare the prediction effects between The Decision Tree and Light GBM models.After researching,we believe that the tobacco manufacturing department should pay full attention to the operation and management level in the relevant production process to ensure the stability of important working conditions,so as to improve the production quality of cigarettes.We also believe that the statistical analysis method of big data will play a essential role in the control and prediction of cigarette machine production process,and it also has important reference value for guiding production.