Influencing Factors And Prediction Analysis Of Tensile Strength Of Living Paper Based On Statistical Method

The paper industry is one of the basic industries of the national economy,and household paper is one of the most common and most frequently used paper products.Tensile strength is used to characterize the maximum tension that paper can withstand under certain conditions and is an important indicator of paper quality.To save costs and reduce energy consumption,In order to save costs and reduce energy consumption,it is necessary to analyse the key factors influencing tensile strength and to develop a predictive model to quantify and evaluate tensile strength.Generally speaking,the analysis of paper tensile strength is mainly characterized by the use of finished paper sampling data,ignoring the influence of some production chain factors.In fact,however,factors in the production chain,such as the operation of machinery,can affect the characteristics of the sampling data and thus the tensile strength.In this dissertation,the operating parameters of paper-making machinery are included in the analysis,and the characteristics of the machine operating parameters and the finished paper sampling data are combined to make an in-depth analysis of the factors influencing the tensile strength of paper and to construct a prediction model for the tensile strength of household paper.The data in this dissertation was derived from paper process data provided by SinoSingapore International Joint Research Institute and Guangzhou Boyd Intelligent Information Technology Co.We first pre-processed the raw data containing the operating parameters and sampling characteristics of the finished paper to determine outliers,fill in missing values and normalise them.On this basis,we used a hierarchical model and grey correlation analysis to carry out an in-depth analysis of the effect of each characteristic on tensile strength.Based on the mediating effect of the hierarchical model,we confirmed that the relevant indicators such as mechanical operating parameters do have an impact on the relevant quality control indicators of the finished paper.Through grey correlation analysis,we found that the correlation between nine mechanical operating parameters,such as design weight and hourly output,and tensile strength was above 0.92,with a high correlation.This in turn confirms that it is reasonable to integrate the operating parameters and sampling data characteristics for the prediction of tensile strength.Finally,in this dissertation,regression prediction models for tensile strength were constructed using PLS,Lasso regression,ridge regression methods and stepwise regression methods from traditional statistical analysis,as well as linear regression and SVM models from machine learning,respectively,and the best-fitting linear regression model was finally selected.Furthermore,considering that some regression models,such as Lasso regression,have a feature extraction function inherent in the prediction.We further compared the effective features extracted by these models when making predictions on the original data with the results of the previous grey correlation analysis,and concluded that the significant impact features of both were more consistent,further validating the validity of our factor analysis.