Research On Optimization Method Of Process Parameters In Wiped Film Evaporation Process

With the popularity of natural materials and the rise of the global trend of returning to nature,especially with the approaching of modernization and internationalization of traditional Chinese medicine,molecular distillation technology has developed rapidly in the separation of high-boiling,heat-sensitive natural substances,At present,molecular distillation technology has been widely used in petroleum,pharmaceutical,chemical and other industries.The wiped film molecular distiller is the third generation molecular distiller and is currently the most widely used.Due to the complex physical,chemical,biological,and material energy conversion and transmission of the wiped film molecular distiller,the system presents severe complexity,uncertainty,and nonlinearity,and it is unable to obtain key variables in time,which makes the modeling of wiped film evaporation process quite difficult,At the same time,for different separation and extraction of objects and external working conditions,the setting of process parameters greatly affects the acquisition of products.In this paper,the following researches are mainly carried out for the above problems:Firstly,according to the characteristics of the membrane model molecular distiller that can not directly establish the mechanism model in the evaporation process,the four process parameters of feed rate,evaporation temperature,vacuum degree and scraper motor speed are selected as the input variables of soft measurement data modeling by soft measurement technology,yield and purity are used as output variables.A soft-measurement model of the wiped film evaporation process was established using the Online Sequential Extreme Learning Machine(OS-ELM).And through comparison with the Support Vector Machine(SVM)and BP Propagation Neural Network(BPNN)simulation results,the results show that the OS-ELM prediction results are more accurate,the running time is shorter,And with the update of the data sample,the model can be continually corrected,which can better reflect the actual operation of the system.Secondly,based on the establishment of the soft-measurement model of the wiped film evaporation process in Chapter 2,The Action Dependent Heuristic Dynamic Programming(ADHDP)algorithm is used to optimize the process parameters of the wiped film evaporation process,and improved the traditional ADHDP algorithm using BPNN to construct critic and action network.By comparing the simulation experiment with BPNN-based Heuristic Dynamic Programming(HDP)and ADHDP algorithm,Theimproved ADHDP has faster convergence speed and higher stability,and the optimal process parameters are better,and the precise control and optimal control of the equipment are realized.Finally,QT software is used to design the interactive graphical user interface(GUI)of the wiper evaporation process parameter optimization system,which mainly includes login,menu,parameter monitoring,model training and parameter optimization functions.It is used for human-computer interaction between field operators and control systems to realize modeling and optimization of process parameters.