Research Of Parameters Measurement And Prediction And Optimzation In Molecular Distillation Based On GA-BP

Molecular distillation is a special liquid-liquid separation technique under high vacuum conditions, which has low temperature, high vacuum degree, short time materials heating and high separation degree etc, meanwhile, the analytes are irretrievable and there is not bulge bubble phenomenon, especially perfect for high boiling points, thermol ability high viscosity and easy of oxidation. It has been widely used in many fields, including the extraction and separation of the active constituents from the Chinese traditional medicine, petrochemical industry, food industry, cosmetics and agriculture.In this paper, during the multi-stage distributed control system (DCS) of molecular distillation experiment, ructus schisandrae crude oil is used to eliminate the terpenoids and alkenes compounds, which is obtained through the supercritical CO2extraction technology. The relationship between process parameters and the product index is graphically illustrated, which provides a superb discussion of these experimental results. In the article, the mathematical model that is about of process parameter of molecular distillation and purity and yield is built, and the forecasting model of product purity and yield is built, meanwhile, optimized model of molecular distillation conditions is built based on artificial neural network and genetic algorithm. There are so much function that one needs to have a software package to make the online process monitoring system and parameter display system to best use of them. The central control subsystem function is perfectly refleted. The main work and research results are shown in the following:(1) Analysis and study of the principle of molecular distillation technology, equipments and application. The experiment is designed by using the scraping membrane molecular distillation device combined with main evaluation index of effecting molecular distillation, which are the evaporation rate and separation efficiency. This article focuses on the issues of relationship between parameters and index based on the experimental data.(2) The fundamentals and training process of BP neural network were clarified. The article proposed improved measures aiming at the limitations and problems. According to the complexity of molecular distillation process and the forecast model was found based on neural network and genetic algorithm,which make use of the characteristic both powerful search ability on the global solution space of the genetic algorithm and strong learning ability of artificial neural network model. On the basis of predicting models, the optimization model of process parameters was set up based on Multi-objective genetic algorithm and the yield and purity are weighted to define objective function. An improved strategy was discussed for firms. (3) System monitoring interface based on GA-BP can be achieved by JAVA technology according to level3molecular distillation systems. This monitoring interface has synchronization function including real-time display of parameters, results of prediction and optimization and identification of the ingredients.