Modeling And Size Distribution Optimization Of The Growth Process Of L-glutamic Acid Crystals

This dissertation takes L-glutamic acid(LGA)as the research object to study its changes in the crystallization process.The kinetic modeling of LGA crystals,the design state estimation and the optimization of the crystallization process control are carried out.First,based on the population balance equation(PBM)of the crystal growth process,the kinetic modeling content of the growth size-dependent LGA was studied to clarify the relationship between the crystal growth rate and solution super-saturation and crystal size distribution Connection.There is a problem of poor numerical stability for the existing Quadrature Method of Moments(QMOM)for solving the population balance model of the industrial crystallization process.In this paper,an automatic differential algorithm(AD)based on Taylor expansion is used to improve the numerical calculation accuracy of the quadrature method.By adding an additional variable set to the original variable set,the quadrature moment of the quadrature is obtained,thereby obtaining a closed solution of the algebraic equation.On this basis,two algorithms,Augmented ODE and Augmented AD-QMOM,are proposed.A numerical example study shows that these two methods can effectively improve the robustness of numerical calculation and significantly improve the accuracy of the moment of solution.In contrast,Augmented AD-QMOM is superior to Augmented ODE in terms of calculation time and accuracy.Then,based on the size-dependent LGA crystal growth process,an augmented extended Kalman filter method(Augmented EKF)based on automatic differentiation method is proposed to estimate the moment state information of the crystallization process.This method abandons the traditional derivation method of EKF and uses automatic differentiation based on Taylor expansion for derivation,so that the derivation speed and derivation accuracy can be optimized.Through the LGA growth process simulation case,it is verified that the Augmented EKF method can be effectively used for the state estimation of crystal growth in the LGA cooling crystallization process,which not only reduces the estimation time,but also improves the estimation accuracy.Finally,based on the above crystallization process moment estimation,a method for optimizing the size distribution of the LGA crystal growth process is proposed.Based on the traditional optimization method of crystal size distribution,this method introduces the estimated size distribution obtained by the above-mentioned moment estimation result through nonlinear inversion technique as a constraint.Simulation results show that this method corresponds to the optimization method given in the recent literature,and significantly improves the fitting degree of the crystal product size distribution to the desired target and optimization time.Experimental results show that new method of this thesis can obtain better product size distribution.