ANFIS Model-based Predictive Control For Pr/Nd Cascade Extraction Process

Rare earth elements are the critical material to develop new kinds of highly sophisticatedproduct for their peculiar properties. However, every rare earth element has its own characters,firstly we should separate them in order to make full utilization of them. Most of theenterprise in the rare earth industry of our country adopt the solvent extraction to separate andpurify rare earth through manual operation, but under this kind of operation we can’tempirically ensure the stability of product for complicated process. It requires to equip withreal time automatic control system to stabilize the quality of products and eliminate theuncertainty of artificial operation.This paper based on the dynamic response curves of the extraction process, establishedan ANFIS model which can describe well of the process. And based on the ANFIS model, amodel predictive controller is designed to regulate the distribution of component content.Concrete researches are as follows:1、By taking the Pr/Nd cascade extraction process as anexample, based on the funnel method, we applied the relevant parameters of an actualextraction process to simulate the extraction process from start to balance, and then obtain agroup of strong non-linear curves and data.2、 According to the dynamic response curves, we established an ANFIS model todescribe the nonlinear dynamic process of Pr/Nd cascade extraction process. The simulationresults show that the established ANFSI model can simulate the nonlinear Pr/Nd extractionprocess at a relatively high precision.3、Based on the ANFIS model, we proposed a adaptive GPC algorithm to ensure thequality of the products. The controller can adjust the flows to keep the component content ofmonitoring states been the standards. And compared with the traditional PID controller, itshows that the GPC controller make a better performance on the component contentdistribution regulation.