Comparative Research Of Optimization Strategies On Selection Of Papermaking Raw Materials And Process Control

Papermaking production is a process chemical industry,whose process simulation and optimization technology are relatively mature at present,however,the research is on the rise about the papermaking raw material selection in the early production process,the rapid analysis technology that guarantees the quality of process control and the controller intelligent optimization design.So the papermaking raw material selection in the production process,the rapid analysis model optimization of pulp chemical properties based on near infrared spectroscopy,the optimization setting of the headbox pressure controller in paper section and the black liquor level controller in Alkali recovery process are selected as research targets.The researches compare the application effects of different optimization strategies according to the optimization goals of different research objects and the corresponding optimization methods,which are aimed to facilitate future use in engineering practice.The main research results are shown as follows:(1)Using MLR and PLSR methods respectively to study the effects of physical properties and chemical composition on the yield of chemical pulping and bio-pulping,which is in order to provide the basis for the optimization selection of the raw materials for the two types of pulping methods.The results of the MLR study showed that:the chemical composition indices had a significant effect on the yield of chemical pulping,and the physical properties had a significant effect on the yield of bio-pulping.However,when establishing the prediction models between physical properties or chemical indices and pulp yield,PLSR could overcome the multicollinearity effects between the performance indices of pulp better than MLR,and obtain an optimized model with a more accurate prediction and reasonable explanation than MLR.(2)In order to realize the rapid determination and real-time analysis for the contents of lignin and holocellulose in pulpwood,the research models of five kinds of pulpwood were established by using near infrared spectroscopy and che mo metrics.In view of the near infrared spectroscopy data of 87 samples,the original spectra were pretreated by multiplicative scattering correction(MSC).and the prediction models of the contents of lignin and holocellulose in pulpwood were established by separately using partial least squares regression(PLSR),principal component analysis and BP artificial neural networks(PCA+ANN),partial least squares and BP artificial neural networks(PLS+ANN).The predicted results showed that PLS+ANN had better predictive performance and higher accuracy,which could overcome the nonlinear problem of near-infrared spectroscopy.(3)The progress of pulp and paper making is very complex with suchcharacteristics as nonlinear,large delay,parameter time-varying,model uncertainty and so on.The traditional PID method has low accuracy in controlling objects while the modern optimization algorithm can optimize PID parameters so as to improve the accuracy,but it still has drawbacks such as slow convergence speed,premature issue and long tuning time.Aiming at these problems,the improved modern optimization algorithms were adopted to optimize process control.In the investigation two modern optimization algorithms were selected:genetic algorithm and particle swarm optimization algorithm.In order to enhance the global optimization ability and convergence rate of the algorithms,the genetic algorithm was improved firstly by optimizing the operators of crossover and mutation,taking the optimal tracking strategy and adding new convergence criterion.The particle swarm optimization algorithm also had some improvements,such as the dynamical adjustment of inertia factor,acceleration factor and the improved convergence criterion.Through comparing the control effects of the headbox pressure in pulping section and the black liquor level in Alkali recovery process with Z-N method,GA,PSO,improved GA and improved PSO,the simulation results in Matlab/Simulink platform showed that the convergence speed of PSO was obviously rapider than GA,and PSO had also higher control accuracy;In addition,the improved genetic algorithm and particle swarm optimization algorithm had faster response speed and better robustness than the Z-N tuning and traditional optimization algorithm,which indicated a high enhance on controlling quality.