Component Content Control For Rare Earth Extraction Process

Rare earths(REs)are non-renewable strategic resources,which are known as "industrial monosodium glutamate" and "the mother of new materials",consequently,they are used in the economic and social development increasingly widespread.China's rare earth reserves and output ranked first in the world.At present,solvent extraction method is adopted by most of the domestic rare earth enterprises for rare earth extraction and separation,however,off-line analysis,experience control,and manual operation mode seriously affect the production efficiency and batch product's stability.Therefore,it is urgently needed for China's rare earth industry to have an automation control production line for rare earth production so as to achieve the transformation of China's rare earth industry,and enhance the competitiveness of enterprises.In this paper,based on the component content actual control for rare earth extraction process,and combined with the actual operation of the site,the "separating funnel method" model is used to simulate the dynamic distribution of component content in the tank of the extraction process to modify the field operation data.The mathematical model for rare earth extraction process is established by data driven method,and the controller is designed based on the model.The specific research contents are as follows:1?Taking CePr/Nd extraction line of rare earth enterprise as a research subject,the parameters of CePr/Nd extraction process are designed according to the cascade extraction theory.The model data of the "separating funnel method" that meets the actual production requirements is used to modify the data collected from the site to ensure the authenticity and completeness of modeling data;2?According to the characteristics of non-linearity of rare earth extraction process,and the difficulty of establishing mathematical model,this paper uses a multi-model method which can linearize nonlinear processes,and replace multiple nonlinear processes with multiple linear sub-models.Firstly,on the bases of the characteristics of sample data in rare earth extraction process,the subtractive clustering algorithm is employed to classify the modeling data,and the number of models is obtained.The recursive least squares method is used to identify the sub-model parameters.Then,the multi-model switching strategy and generalized predictive controller are designed according to the built model.Finally,it summarizes the control law under different disturbance conditions by simulation experiment.3?As the component content of the monitoring points in the field of rare earth extraction is variable in a restricted regions,the component content zone control method is put forward.Because of the instability of the multi-model switching in the multi-model modeling process,the echo state neural network algorithm is adopted to model the rare earth extraction process,which has a capability of nonlinear approximation and can solve the memory fade problem in traditional network.According to the network model and the characteristics of CePr/Nd extraction process,the zone control strategy of CePr/Nd extraction process is set up,and the component content zone controller is designed based on the generalized predictive control algorithm.Simulation tests with the field data are presented to show the validity of the proposed method.In this paper,the modeling and control method of the rare earth extraction process are studied by using the multi-model method and the zone control method.The experimental results show that the control rules are in accordance with the field operation rules under the condition of the extraction conditions is effected by various disturbances,and it has a certain guiding significance for adjusting diverse flow rate in the rare earth extraction site.