Research On The Optimization Method For Parameters Of Molecular Distillation Process

Molecular distillation is a new and special kind of separation technique, which has the characters that the distillation temperature is low, the heating time of materials is short, the process is conducted under high vacuum situation and at the same time there is no boiling phenomenon, etc. These characteristics make it very suitable for the separation of heat-sensitive, high boiling point and easily oxidized substances, And presently it has been widely used in the pharmaceutical industry, food industry, petrochemical industry and agriculture industry, etc.Firstly, the molecular distillation experimental platform, which the evaporation temperature, rotor speed and the feeding rate are adjustable, is constructed. And then the experiment is carried out with the fructus schisandrae crude oil obtained by supercritical fluid extraction as raw materials. Regarding to the factors of distillation temperature, feed rate and rotor speed, the single factor experiment method is adopted in order to get the experiment data of the factor’s influence on the product purity and yield rate. Then a comprehensive analysis on these data is carried out, which lays the groundwork for the subsequent research of multi-objective optimization.Generally, the measurement index are more than one, so there are always problems with the molecular distillation process parameters based on the traditional single objective optimization in guiding the actual production. In order to solve the problems of traditional molecular distillation parameter optimization method, establishing a multi-objective optimization model of molecular distillation process is proposed in this paper. With the experimental data of the molecular distillation process parameters and product purity and yield rate, the function model of the product purity and yield rate are established respectively based on RSREG modeling principle, and the molecular distillation process parameter optimization problem, in fact, is transformed into a multi-objective optimization problem. Then the multi-objective genetic algorithms, NSGA Ⅱ, is adopted to optimize the established optimization models of molecular distillation process. Finally, the most appropriate parameters combination would be chosen according to the actual production needs from the set of Pareto optimal solution, and then the selected parameters combination would be tested in actual production. The experimental results of the fourth and first groups of parameters show that the error of purity between the actual value and the predicted value are about 0.74 percent and 1.1 percent, and the error of yield rate between the actual value and the predicted value are about 0.55 percent and 0.6 percent.The accuracy of established multi-objective optimization model for molecular distillationand the feasibility of using NSGA Ⅱ to optimize multi-objective optimization model are verified, which gives us a feasible method to get molecular distillation process parameters of more practical guiding significance to the actual production.