Optimal Control Research And Application Of Scratch Type Molecular Distillation Process

Molecular distillation is a new separation technique,which separates the liquid mixtures under the high vacuum and has the characteristics of low distillation temperature(lower than the boiling point of material),short heating time,high separation degree and so on.Also this separation process is irreversible,and no bubbling phenomenon.This technique is especially suitable for the separation of high boiling point,high viscosity,heat sensitive and easily oxidized substances.At present,the technology has been widely used in the pharmaceutical industry,such as the extraction of effective components of vitamins and herbs,the petrochemical industry,food industry,cosmetic industry and agriculture.In the research filed of molecular distillation,the researching about the modeling,optimization of process parameters and the control algorithm of the control volume is few.However,these studies play an important role in the energy saving of production,improving the effective production time and reducing the dependence on artificial experience.So the relationship between process parameters and product parameters is discussed according to the results of purification of crude oil from Schisandra chinensis in this pape,and by the theoretical analysis of molecular distillation,the main parameters affecting the purity and yield of the purified product can be obtained.On this basis,the modeling,multi parameter optimization and control algorithm for the process of molecular distillation are studied in this paper.The main work and research contents of this paper are summarized as follows:Firstly,a prediction model based on Extreme Learning Machine for is proposed for the process of scraping film molecular distillation.Due to the characteristics of nonlinear,strong coupling and large delay of molecular distillation system,it is difficult to realize the mechanism modeling,and taking into account the local optimization of BP network,the shortcomings of long training time,a prediction model of molecular distillation system based on extreme learning machine is proposed.The simulation results show that the prediction model can predict the output state of the system more accurately and timely.Secondly,the Heuristic Dynamic Programming method is used to optimize the parameters of the wiped film molecular distillation process.Because of the coupling between parameters,the optimization effect of single parameter is not ideal,and several parameters need to be optimized many times.By this method proposed in this paper,we can find the better process parameters in any initial state and shorten the time of parameter adjustment.If the BP network is used to realize the Heuristic Dynamic Programming,the problems such as the local optimum and the long time of BP network will also be introduced into the algorithm,so a new method that taking Extreme Learning Machine to implement the Heuristic Dynamic Programming is proposed in this paper and give the theoretical derivation.This method will improve the speed of the improved algorithm nearly doubled.Thirdly,the inverse model control method of nonlinear system based on Online Sequential Extreme Learning Machine is proposed.This method not only solves the problem that the analytic inverse system is difficult to obtain,but also realizes the on-line adjustment of the inverse system.In this method,the control of the nonlinear system can be equivalent to the control of the first order system by the series of the inverse system and the original system.The proposed method can effectively control the amount of the molecular distillation system such as the motor speed and the temperature control of the hot oil pump.Finally,this paper designs the control scheme of the field control bus for the film scraping molecular distillation,which makes the automation degree of the molecular distillation system further improved,and the reliability and maintainability are also improved.And by using of OPC communication technology,the results of the advanced algorithm running in the upper computer can be sent to the advanced algorithm of the results of the scene of the controller through PROFIBUS network.