Model Predictive Decoupling Control For Rare Earth Extraction Process

With the development of rare earth industry in the direction of large-scale,centralization and continuation,it is an inevitable trend in the future to establish an industrial line with high efficiency,stability and intelligence.In the actual field,due to the interference of external environment and internal factors of extraction industrial system,it is necessary to effectively control the key quantity of the whole system to ensure the quality and benefit of rare earth products.The automation level of rare earth extraction industry in China is low,and the extraction industry line relies on manual experience to control the flow rate,which leads to the instability of product quality and the waste of raw material resources.Therefore,it is very important to select effective automatic control methods to ensure the stable and efficient operation.The main object of this paper is the extraction and separation process of CePr/Nd.Firstly,the separation funnel method is used to simulate the process by computer,and the operation data is obtained to supplement and modify the data collected in the field.Aiming at the strong coupling of the extraction process and the complex and changeable environment background,the system model is established according to the idea of data-driven,and the appropriate optimization controller is designed to ensure the stability of products quality.The contents are as follows:1.To solve the problem of insufficient data collected in the field of rare earth extraction,the technological parameters were set based on the theory of cascade extraction.The computer dynamic simulation of CePr/Nd cascade extraction system is carried out by using the method of liquid separation funnel,and the separation coefficients of multi-components in each stage are calculated dynamically.The data of operation process are obtained to supple and modify the data collected on site,which lays a foundation for the modeling and control of rare earth extraction process.2.Continuous disturbances caused by environmental changes,equipment damage and batches of raw materials in rare earth extraction process will lead to changes in operating characteristics of objects,and resulting in off-line model mismatch and deterioration of controller performance.Based on the off-line echo state network model of component content,an on-line optimization control method of rare earth extraction process model is proposed.When the output error exceeds a certain threshold,the model optimization adjustment strategy is started.The model parameters are corrected by Kalman filtering,and the predictive controller is corrected accordingly.Finally,the feasibility design is verified by experimentalsimulation.3.In view of the strong coupling of rare earth extraction process,this paper regards the system as multiple multiple-input and single-output subsystems,and constructs the extreme learning machine model for each subsystem.On this basis,the decoupling control strategy is incorporated into the predictive control algorithm to design the decoupling model predictive controller,so that the deviation weight of the objective function of a loop controller would be adjusted adaptively according to the output deviation of other loops.Finally,the design is compared with the conventional generalized predictive controller,and the control regulation laws under different conditions are summarized and the reliability of the controller is verified.To improve the operation and control performance of rare earth extraction process,based on establishing appropriate system model,the model online optimization controller and decoupling model predictive controller are designed respectively,and the reliability of two controllers is verified by simulation.This provides certain theoretical basis and technical support for intelligent control system and a more stable application to the extraction process places with changeable environment.