Study On Technology Optimization Control Of Wiped Film Molecular Distillation

Molecular distillation,which also defined as short path distillation,is a special separation technology of low temperature(relative to boiling point),non-equilibrium and continual distillation process at high vacuum condition.Molecular distillation has achieved unprecedented development due to the advantages unmatched by conventional distillation,which has played a unique role in high-boiling,heat-sensitive,and easily oxidized substances.The wiped-film molecular distillation technology is more commonly used in experiments and actual production due to its unique structure and separation advantages.Because the molecular distillation has a complex system with many variables,strong coupling and nonlinearity,it is too difficult to establish an accurate model to achieve the effective control.So the determination of process parameters depends on the specific operating experience and test Operation.It is great of studying the optimization parameters of the wiped-film molecular distillation by the intelligent optimization algorithm.In this paper,several major parameters affecting the purity and yield of the purified material were obtained through the theoretical analysis of molecular distillation,and combined with the MDS—80—II molecular distillation device of the second stage of the molecular distillation device of Guangzhou Hao Li biological science and Technology Co.,Ltd.The following research was conducted to purify experimental data of Schisandra crude oil:First of all,the model network of Dual Heuristic Dynamic Programming(DHP)algorithm based on BP neural network(BP-DHP)is used to Establish a predictive model of wiped-film molecular distillation,and verify by simulation that it can accurately predict the purity and purification yield at the next moment;Secondly,due to the non-linearity,strong coupling and complicated process of the scraped-film molecular distillation process,and then take the BP-DHP algorithm to simultaneously optimize the four main parameters of scraping membrane effect wiped-film molecular distillation,and the simulation results of BP-DHP algorithm are Compared with the optimized results of the Heuristic Dynamic Programming(HDP)(BP-HDP)algorithm achieved by the BP neural network,all of them can reach a stable and superior value.Furthermore,in order to reduce the parameter adjustment step,we take advantage of the characteristic of the speed of Extreme Learning Machine(ELM),and using the Moore-Penrose generalized inverse matrix idea of ELM to adjust the weight matrix of the DHP algorithm model network,the critic network and the action network The ELM-DHP algorithm was proposed to reduce the number of adjustment steps and quickly give better parameters to improve the purification effect.Finally,in order to improve the stability of the rotational speed,a double close-loop control method of inner ring current hysteresis loop and outer ring ant colony algorithm(Ant Colony OptiMization,ACO for short)optimization PID(ACO-PID)is used to control,so that the wiped film motor maintains a relatively stable speed at low speeds.