Research On Dynamic Optimization And Model Predictive Control Of Azeotropic Distillation Process

Azeotropes,such as ethanol-water and acetonitrile-water,are widely present in chemical production processes,and it is difficult to effectively separate azeotropes by common distillation methods,while azeotropic distillation technology can achieve the effective separation and purification of multiple azeotropes.Azeotropic distillation is more difficult to be controlled than other distillation processes,such as extractive distillation,due to the presence of azeotropes and the limitations of azeotropic composition and harsh operating conditions.In addition,in order to save energy and reduce consumption,energy-saving technologies such as heat integration and heat pumps are usually used,which often increase the coupling between azeotropic distillation variables,making it difficult to control azeotropic distillation with heatintegrated.In this paper,the characteristics and control difficulties of azeotropic distillation process are systematically analyzed,and the research of PID control scheme design and model prediction control method is carried out based on typical binary azeotropic system of acetonitrile-water and ternary azeotropic system of isopropyl alcohol-diisopropyl ether-water.The main research contents of this paper include:（1）Using the slope criterion method to determine the temperature sensitive stages of acetonitrile-water azeotropic distillation and isopropanol-diisopropyl ether-water azeotropic distillation,studying the relationship between acetonitrile-water azeotropic makeup and product purity,designing two feasible temperature control schemes based on PID control,and realizing the stable operation of the two processes and the control of product purity;Based on the temperature control scheme,the effective variables for manipulating the product purity were analyzed,and the control scheme was improved by setting a suitable composition controller to achieve high precision and robust control of the product purity.The results show that the proposed control scheme achieves stable and robust operation of the process under ±10% and ±20% feed flow and composition disturbance,respectively,and ensures the product purity.Among them,the maximum dynamic deviation,response time and steady-state deviation of the control system are reduced by the optimized control of the makeup,and the dynamic performance of the control system is improved.（2）Based on the PID control scheme,a Model predictive control（MPC）method based on a linear time-invariant state space model is designed,and a multi-objective optimization of MPC controller weights is performed by genetic algorithm,and a combined PID-MPC control method is proposed.In the azeotropic distillation process of acetonitrile-water,the MPC controller replaces the original temperature control,integrating multiple variables and simplifying the control structure;in the azeotropic distillation process of isopropanol-diisopropyl ether-water,the MPC controller replaces the original composition controller.The combined control scheme was compared and analyzed by qualitative and quantitative analysis.The results show that the MPC controller can effectively control several key variables such as temperature and purity,and its internal prediction model can effectively reflect the coupling relationship between these variables,and the maximum dynamic deviation of MPC control is smaller than that of PID control,and the corresponding Integral of squared error criterion（ISE）value is smaller,which has better stability and robustness.（3）Using the mathematical relationship between easily measurable auxiliary variables and product purity,a prediction model for real-time online prediction of product purity is constructed based on time-delay neural network（TDNN）,and combined with MPC method,a fusion control of artificial neural network-model predictive control（ANN-MPC）method is proposed to achieve high precision control of product purity without direct detection of product purity.The qualitative and quantitative analysis of the control method is carried out by example,and the results show that the ANN-MPC control scheme has no significant increase in the transient maximum dynamic deviation compared with the MPC-PID scheme without using composition measurements,and the product purity can still be controlled with high accuracy,and the corresponding ISE value is close to that of the MPC scheme,and the control effect is better than that of the PID scheme,which has certain advantages in industrial applications.The research in this paper provides a feasible control scheme for the industrial application of azeotropic distillation processes such as acetonitrile-water and isopropanol-diisopropyl ether-water,and explores the advanced control methods for azeotropic distillation processes,which has certain reference value for the research of dynamic optimization and control methods of azeotropic distillation.