Modeling And Intelligent Optimizing Control For Rare Earth Extraction Process

There exists abundant rare-earth resource in mainland of China. And reserve and production of the rare earth occupy the primary position in the world. Rare earth is indispensable industrial raw material in the production of display phosphor, NdFeB permanent magnet, Ni/MH battery, rare-earth aluminum alloy, precision ceramics etc. The extraction technics begins with the material liquid confection by dissolving the rare-earth extractive mine at a certain composition ratio. Using the feeding equipment, the material liquid is affluxed into the middle stage mixer-settler. At the same time solvent from the first stage and acid from the last stage affluxed into the extraction process. Via stage-wise extraction, hard-extracted and easy-extracted components are exported from aqueous and organic exits. The hard-extracted and easy-extracted components are important quality index of the rare-earth extraction process. So under the condition of maintaining production and cost, it is important to improve the element component content (ECC), i.e. quality of rare-earth product.ECC of hard-extracted or easy-extracted component is hard to be measured online. There exit strong coupling, nonlinear, large time delay between ECC and solvent flowrate, material liquid flowrate, acid flowrate, and the relation can not be described by precise model. ECC also varies with the disturbances of solvent saponification degree and feed-in compositions. So it is difficult to adopt the existing control methods to control the ECC. The rare-earth extraction process is still under manual control. The stability and consistency of product's quality can not be ensured.Under the support of the Key Technological Project "industrialization technology of on-line analysis and closed-loop control for the rare earth extraction process" (2002BA315A-4) in 10th five year plan, aiming at improving the product's quality, modeling and intelligent optimizing control for rare earth extraction process is developed and acquired the following achievements:An intelligent optimizing control method is proposed. And In order to validate the proposed optimizing control method, a dynamic model for rare-earth extraction process is proposed. An optimizing control simulation experimental system for rare-earth extraction process is designed and developed based on the extraction dynamic model, actuating mechanism and detecting virtual device, PLC, monitoring and optimizing software etc. Some relevant simulation experiments are conducted on the simulation experimental system. The results validate the effectiveness of the proposed optimizing control method. The main works are as follows:(1) The proposed intelligent optimizing control method to control the ECC within required range is composed of solvent, material liquid, acid flowrate setting layer and the above flowrates'loop control layer. The setting layer consists of initial setting model of olvent, material liquid, acid flowrates, ECC soft sensor model,, solvent flowrate compensation model and acid flowrate compensation model. The initial setting model adopts the technics equations and the case based reasoning (CBR) technology to give the initial settings of solvent, material liquid and acid flowrates. The compensation terms of solvent and acid flowrates are educed by fuzzy reasoning technology, using ECC errors of hard-extracted components between offline analyzed value, soft-sensor value and setting values as the solvent compensation model's input, ECC errors of easy-extracted components between offline analyzed value, soft-sensor value and setting values as the acid compensation model's input.(2) A P507(Kerosene)-HCL separating LaCe/PrNd extraction system is considered, which uses the mixer-settlers as the main apparatus. The stagewise dynamic model for single component is combined with separation factors of multiple component balance derived from "Counter Current Extraction Theory". A high-order nonlinear dynamic model of stagewise extraction with backmixing is proposed. The concentrations of all elements in aqueous and organic phase at different stages are regarded as state, the solvent, material liquid and acid flowrates as input, ECC of hard-extracted and easy-extracted components as output. In order to validate the model, the static state properties are simulated firstly. It is implemented by keeping the flowrates of solvent, material liquid and acid as constants, and then the concentration changing from initial condition to a new static state is analyzed and compared with measured data.Secondly, the influence of step changes of solvent, material liquid and acid flowrates on the element concentrations is tested. The simulation results show that the dynamic model can simulate the dynamics of the extraction process.(3) The soft-sensor model in the intelligent optimizing control method is based on extraction dynamic model. The multiple linear models at multiple running points are used as the principle model, and a neural network as the error compensation model. Measured data of hard-extracted component B(La, Ce) and easy-extracted component A(Pr, Nd) is acquired form a P507(Kerosene)-HCL separating LaCe/PrNd production line. The above data is used to validate the soft-sensor model. Experiment results indicate that the precision of the hybrid intelligent soft-sensor is relatively high.(4) The simulation experimental system for rare-earth extraction process optimizing control is composed of virtual plant software based on the dynamic model, intelligent optimizing control software based on the proposed intelligent optimizing control method, virtual plant computer, monitoring and optimizing computer, Siemens S7-400 PLC, actuating mechanism and detecting virtual device etc. For a P507(Kerosene)-HCL separating LaCe/PrNd production line, the validity of the intelligent optimizing control method is tested by some experiments which involving production changing, feed-in composition changing, saponification degree changing. The results show that the ECC can be controlled within the index range even when the production is changing, composition and saponification degree are fluctuating.