Study On Concentration Control Of Rare Earth Oxide In Rare Earth Molten Salt Electrolysis

In rare earth electrolysis process,there are many complex physical and chemical changes in the electrolytic cell,Many parameters in electrolytic cell are uncertain and cannot be measured continuously or online,which is a complex industrial process.Rare earth oxide is the raw material of rare earth electrolysis.The concentration of rare earth oxide in the electrolytic cell is related to the material balance in the electrolytic cell during the production process,and will directly affect the quality of rare earth metals produced.At present,most domestic and foreign rare earth electrolytic manufacturers still use manual way to add rare earth oxide,a few manufacturers use fixed time and fixed point feeding device to add rare earth oxide,these feeding methods do not fully consider the influence of the concentration of rare earth oxide inside the electrolytic cell on the electrolytic process.After in-depth analysis of rare earth electrolysis process,this thesis made full use of the data collected on site,combined with BP neural network,evolutionary algorithm,fuzzy control and other theoretical knowledge,the control of rare earth oxide concentration in the electrolytic cell during rare earth electrolysis process was studied,MATLAB/Simulink software was used to complete the simulation and verification.The main research contents are as follows:In view of the current situation that the concentration of rare-earth oxide cannot be directly measured,the relationship between the cell resistance and the concentration of rare-earth oxide is qualitatively analyzed by referring to the correlative control idea in the field of aluminum electrolysis,and the concentration of rare-earth oxide in the electrolytic cell is indirectly monitored by tracking the cell resistance and the change rate of the cell resistance.Therefore,the prediction method of groove resistance is studied and the prediction model of groove resistance based on BP neural network is constructed.The data collected from the 8000 A neodymium electrolytic cell in a rare earth plant in Baotou,Inner Mongolia were used to calculate the cell resistance.The data were preprocessed to analyze the timing characteristics of the cell resistance,and the nonlinear mapping ability of BP neural network was used to predict the cell resistance.In order to improve the prediction accuracy,the weight and threshold of BP neural network were optimized by thinking evolution algorithm.Compared with the actual slot resistance data,the prediction accuracy of the model for slot resistance was±0.3μΩ,which met the requirement of setting error accuracy.The relationship among rare earth oxide concentration,tank resistance and blanking time interval was analyzed.The concentration was monitored by tank resistance,and the concentration was controlled by adjusting the blanking interval.The fuzzy controller of Mamdani rare earth oxide concentration was designed.Step response method is used to complete the identification of rare earth oxide concentration model in rare earth electrolytic cell,and then complete the simulation of rare earth oxide concentration fuzzy control system,and compared with the traditional PID control system,the simulation results show that the fuzzy controller designed in this paper has good robustness to control the concentration of rare earth oxide.The dissolution of rare earth oxide in electrolyte molten salt is a nonlinear process,and there is a dissolution lag.In order to further improve the control performance of the system,the neural network prediction module and fuzzy control module are combined to establish a fuzzy control system of rare earth oxide concentration based on neural network prediction.After simulation verification,The rare-earth oxide concentration fuzzy control system with neural network prediction link has stronger anti-interference ability,better stability and faster response speed than the fuzzy control system alone.