Development And Application Of Best Practice Model For Energy Consumption Of Tissue Paper Machine Based On Data Mining Methods

The tissue papermaking process consumes a lot of energy.How to reduce the energy cost in the tissue paper production process without affecting the quality and output of tissue paper is an urgent problem to be solved.The papermaking process is a highly automated process.A large amount of historical data has been accumulated during the production process,which contains a lot of information.How to dig out potential patterns and rules from these large amounts of data into the knowledge base,reduce the energy cost of the entire production process,and provide guidance for the lean operation of enterprises,has been highly valued by academia and industry.Regarding the parameter adjustment of household paper machines,most paper companies rely on the experience of front-line workers.However,the optimization of the papermaking process is a multi-parameter optimization process,and each parameter is coupled with each other,while taking into account the quality and cost of the base paper,which cannot be achieved by the experience of workers.Starting from the historical data of the factory,this research collects the production process parameter data and quality data of a tissue paper company in2020.The data is preprocessed in combination with the actual production situation,and abnormal data is eliminated.Using the pre-processed data,the contribution of steam and electricity to energy costs is analyzed.In 2020,the cost of high-pressure steam will account for49.74% of the total production cost.Subsequently,the impact of product quantification and winding speed on the energy cost of a unit product is analyzed.The energy cost per unit product of the same product fluctuates up to 100 yuan/ton.The unit product cost of this tissue paper company fluctuates greatly,and there is a large room for cost optimization.In order to optimize the energy cost of tissue paper machine through parameter adjustment,this research first uses regression algorithm to regress and fit process parameters and unit product cost.Among them,the correlation index and average relative error of random forest algorithm are 0.98 and 0.41%,respectively.Compared with AdaBoost algorithm,KNN algorithm,Lasso algorithm and Ridge algorithm,it has obvious advantages.Secondly,the random forest algorithm is used to sort the importance of features,and the control variable method is adopted to eliminate the features with low importance in turn,and the parameters that need to be adjusted are determined.Then use the historical data to establish the best working condition database,and develop the best practice model for the tissue papermaking process based on the association rule algorithm in the data mining algorithm.Using historical data for verification,the results show that the best practice model can reduce the energy cost of a unit product by about 3%.Finally,by reading real-time production data,60 offline tests were performed on the model,and the average energy cost per unit product was saved by 26.54 yuan/t.By collecting optimized soft-sensing data,it is proved that the energy consumption best practice model will not reduce the quality of the base paper while optimizing the energy cost of the unit product.In order to realize the online development of the best practice of energy consumption,this research realized the online development of the model through the three steps of functional module packaging,industrial preparation and industrial application.The online model interface was developed from the two dimensions of managers and field operators,and finally the online model was tested in industry.The results of the industrial test show that the company has a large energy-saving space for this production line.For a tissue paper manufacturing enterprise with an annual production capacity of 80,000 tons,it can save 1,391,200 yuan in energy costs each year.The online model has functions such as real-time working condition detection,real-time working condition optimization,optimization result output,and working condition data update.The online model can realize the intelligent identification and optimization of working conditions,without the need for on-site operators to perform any operation on the model.The development and application of the online model proves that the best practice model has good industrial application value.