Development Of Online Soft-sensor Model Of Crepe Quality Of Tissue Based On Intelligent Algorithm

Creping is the most important part of the production of tissue paper,which not only affects the production efficiency and post-processing of the papermaking machine,but also directly affects the paper quality.With the improvement of people's living standards,people have higher requirements on the quality of tissue paper and the classification of crepes and the quality control of crepes.At present,the classification of crepes by domestic tissue companies is mainly based on manual visual inspection,and it exist subjective errors.It is not possible to monitor the quality of crepes online in real time,and it has also affected the improvement of the quality of crepes in tissue paper.With the development of the Industrial Internet,papermaking enterprises have accumulated a large amount of historical data.Using data mining and machine learning-related knowledge to mine valuable information in historical data is the direction of the development of the Industrial Internet.This research is aimed at tissue companies and based on historical data of papermaking companies.Through correlation analysis and other methods,the appropriate variables are selected as the input of the online soft measurement model of crepe quality.A soft-sensing model of tissue crepe quality was established to realize online soft-sensing of tissue crepe quality.And this study by preprocessing papermaking process conditions and energy consumption data,and using the kernel principal component analysis(KPCA)to reduce the dimension of data,and then using DBSCAN(Density-Based Spatial Clustering of Applications with Noise)to cluster the data.Finding the internal characteristics between working conditions,and identify the working conditions,finally analysis each type of energy consumption,the cost and other indicators,finding the optimal working conditions,and assess the energy saving potential.Finally,the established model will be developed and industrially verified,so that the model is really put into use and creates value for the enterprise.First of all,in order to establish a soft-sensing model for tissue paper crepes,the paper first measures a variety of technical indicators related to tissue crepes.Afterwards,through correlation analysis,several indicators that have a greater effect on crepe levels were selected as output results.By investigating the influence of various links in paper production on the quality of crepes,a suitable input variable is selected.Finally,based on the selected variables,a soft-sensing model of tissue crepe quality based on a gradient boosting regression tree is established.The results show that the average relative error of the testing data is less than 5%.This model solves the problem of online soft measurement of crepe levels.It provides a new method and basis for the quality control of the tissue production process.Then,in order to solve the problems of large energy consumption and irrational operation parameter setting in the papermaking industry,this paper analyzes the papermaking process energy system,combines the characteristics of multiple working conditions of the creping process,and uses the DBSCAN clustering algorithm to make model of creping process.Identify the different working conditions of the wrinkling process,analyze the energy consumption and cost,and mine the energy saving potential of the production process to achieve the purpose of energy saving and emission reduction.The results show that for the domestic paper mill with an annual output of 80,000 tonnes,the annual energy saving potential about 8.8 million RMB/year based on the energy cost of the place.Finally,based on the establishment of a soft-sensing model of tissue crepe quality,the development and application of a soft-sensing model of tissue crepe quality is realized.After a period of development and testing,the model was successfully used in a large domestic paper company.The results show that the average relative errors of surface roughness,crepe amplitude,and crepe frequency are all within 5%.The model works well and the errors are within an acceptable range.It has a good industrial application prospect,and it is of great significance to improve the quality of tissue paper crepes.