The multiple models predictive control of component content for the rare earth extraction procession

Abstract

Due to the characteristic of rare earth extraction separation, combined with the material balance model, an approach based on multiple model is presented in this paper. Firstly, by using the data selected in an industrial process, the steady points are obtained, which use the improved subtractive clustering algorithm. The recursive least-square identification method is then adopted to identify the model parameters. The product Y can be predicted on-line with high purity in the rare earth extraction separation process, which choosing the best performance index function. And an experiment with real industrial operations data is implemented to verify the proposed method. Finally, general predictive controller corresponded is designed for each sub-model so that component content is controlled real-timely and accurately. Simulation results show the effective performance of the referred method.