Study On Soft Sensor Model Of Low Temperature Emulsion Styrene-butadiene Rubber

As a kind of synthetic rubber,Styrene-butadiene rubber(SBR) has been widely used in our daily life and military areas,due to its good performances,as a substitute for the natural rubber in many aspects.The real-time monitoring of polymerization conversion rate is a difficulty for SBR low temperature emulsion production.At present,laboratory analysis methods are used to control the index in many domestic enterprise.The major defects for these control methods are big time lags,which affect the control effects on the one hand,and consume a lot of resources on the other hand,no matter manpower or material costs,which are relatively high.The development of the soft sensor technology provides a good approach for solving the problems.So far,there are few reports on the study of the polymerization conversion rate of low temperature emulsion SBR by using this technology.In the case of low temperature emulsion SBR production,although the production process is very complex,the working state is relatively stable,there are strong nonlinear relations between auxiliary variables,and the enterprise also has higher accuracy requirements for the prediction of the total conversion rate.In this paper,a soft sensor modeling method is used to accurately predict the polymerization conversion rate of low temperature emulsion SBR.Based on integrated pruning,this paper constructs a soft sensor modeling method.First,bagging method is adopted to establish different LS-SVM weak learners,then AdaBoost.RT was used to trim the weak learners,and the pruned weak learners are weighted and exported.The method proposed in this paper has high prediction accuracy,and solves the problem that the integrated algorithm takes up a lot of resources and theprediction speed is slow.In addition,some problems of LS-SVM,such as sparsity,robustness and so on,are more effectively compensated.The simulation application of the soft sensor modeling method is carried out,and the results show that the method is more suitable for modeling industrial process of low temperature emulsion SBR with stable working conditions and high estimation accuracy.