【作者】 王瑶；

【导师】 刘涛；

【作者基本信息】 大连理工大学 ， 控制理论与控制工程， 2021， 硕士

【摘要】 结晶是化工生产过程提取产物的主要方式之一,通过结晶得到的晶体的质量影响着下游生产。论文主要对L-谷氨酸冷却结晶过程的状态估计算法和操作条件优化进行了研究,目的是提高L-谷氨酸批次生产过程的生产效率和晶体产品质量。研究的模型是粒子数衡算模型。为便于模型求解,采用求积矩量法(Quadrature Method of Moments,QMOM)对该机理模型进行简化,最终研究的模型是矩量形式的粒子数衡算模型。针对矩量形式的粒子数衡算模型,提出了提升的正交求积矩量法(Augmented Quadrature Method of Moments,AQMOM),用来提高数值求解效率。针对结晶过程晶体尺寸分布(Crystal Size Distribution,CSD)难以进行实时测量的问题,设计一个非线性状态估计器。利用对结晶过程溶液浓度的实时测量值,结合矩量模型,提出一种改进的时域长度可变的滚动时域状态估计算法,得到结晶过程CSD的矩量信息。通过将该算法应用于一个非等温化学搅拌槽反应过程案例、以及L-谷氨酸结晶的尺寸依赖生长-成核过程仿真案例(标称情况和不确定情况),验证说明本文提出的状态估计算法相比于扩展卡尔曼滤波(Extended Kalman Filter,EKF)和经典的滚动时域估计器(Moving Horizon Estimation,MHE)具有更好的估计精度和鲁棒性。针对L-谷氨酸冷却结晶过程的操作变量优化问题,采用最优控制理论中的玛雅(Mayer)指标来设计目标函数,通过求解目标函数来确定最优的操作条件。基于结晶过程矩量模型,将期望晶体产品CSD转换为期望的CSD矩量,对于晶体的纯生长过程和生长成核过程分别定义不同的目标函数,建立相应的操作条件优化算法。仿真结果表明,相比于工程上常用的线性和程序降温策略,采用该文提出的优化降温策略可以使晶体生长到期望的CSD并且明显缩短结晶过程操作时间,使晶体生长更为集中。最后,利用L-谷氨酸冷却结晶实验平台,对提出的操作条件优化策略进行了实验验证,并且与线性降温策略和程序降温策略进行比较。采用颗粒分析仪FBRM检测晶体产品的尺寸分布。实验结果表明,提出的优化方法可以提高L-谷氨酸晶体产品的质量,而且显著缩短冷却结晶过程时间。更多还原

【Abstract】 Crystallization is one of the main ways to extract products in the chemical production process,and the quality of crystals obtained through crystallization affects downstream production.The thesis mainly studies the state estimation algorithm and operating condition optimization of L-glutamic acid cooling crystallization process,the purpose is to improve the production efficiency of L-glutamic acid batch production process and the quality of crystal products.The model studied is the particle balance model(PBM).In order to facilitate the solution of the model,the quadrature method of moment(QMOM)is used to simplify the mechanism model.The final research model is the PBM in the form of moments.Aiming at the PBM form of moments,an augmented quadrature method of moments(AQMOM)is proposed to improve the efficiency of numerical solution.Aiming at the problem that the crystal size distribution(CSD)of the crystallization process is difficult to measure in real time,a nonlinear state estimator is designed.Using the real-time measurement value of the solution concentration in the crystallization process,combined with the moment model,an improved rolling time domain state estimation algorithm with variable time domain length is proposed to obtain the moment information of the CSD in the crystallization process.By applying the algorithm to a non-isothermal chemically stirred tank reaction process case and the size-dependent growth-nucleation process simulation case of L-glutamic acid crystals(nominal conditions and uncertain conditions),the state estimation proposed in this paper is verified and explained.Compared with the extended Kalman filter(EKF)and moving horizon estimator(MHE),the algorithm has better estimation accuracy and robustness.Aiming at the optimization problem of operating variables in the cooling crystallization process of L-glutamic acid,the Mayer index in the optimal control theory is used to design the objective function,and the optimal operating conditions are determined by solving the objective function.Based on the crystallization process moment model,the desired crystal product CSD is converted into the desired CSD moment.Different objective functions are defined for the pure crystal growth process and the growth nucleation process,and the corresponding operating condition optimization algorithm is established.The simulation results show that,compared with the linear and program cooling strategies commonly used in engineering,the optimized cooling strategy proposed in this paper can make the crystal grow to the desired CSD and significantly shorten the operation time of the crystallization process,making the crystal growth more concentrated.Finally,using the L-glutamic acid cooling crystallization experimental platform,the proposed operating condition optimization strategy was experimentally verified,and compared with the linear cooling strategy and the program cooling strategy.A particle analyzer(FBRM)was used to detect the size distribution of the crystal product.Experimental results show that the proposed optimization method can improve the quality of L-glutamic acid crystal products and significantly shorten the batch time of cooling crystallization process.更多还原