Research On Predictive Control Technology Of Post-combustion CO₂ Capture System

Under the background of the national policy of actively responding to the greenhouse effect and implementing carbon peak carbon neutrality.The development of Post-Combustion CO₂ Capture（PCC）technology from Coal-fired Power Plant（CFPP）is an important means to reduce CO₂ emission and realize sustainable development in our country.Since MEA solvent is very suitable for treating flue gas with low CO₂ partial pressure in coal-fired power plants,and is highly compatible with existing coal-fired power plants,the post-combustion carbon capture technology based on MEA has become one of the technologies with the highest capture efficiency and the broadest application prospects.With the continuous popularization of this technology,the modeling and optimal control of PCC system has become a hot research topic.The system is a typical multivariable system,which has many complex characteristics such as strong nonlinearity,large lag and strong constraint.It is very important to study the process flow and dynamic change of PCC process deeply,and adopt advanced control strategy to control it,so as to ensure the flexible operation of the system and improve the economic benefit of the factory.Based on the above background,the main research contents of this dissertation include:（1）According to the process flow and chemical reaction process of the system,the complete steady-state and dynamic models were built in Aspen Plus^?and Aspen Plus Dynamics^?software,and the step response experiments were carried out on the main variables,and the dynamic change rules among the main variables of the PCC system were analyzed.At the same time,the gap measurement is used to determine the nonlinear distribution of the model,which provides the model basis for the subsequent controller design.（2）According to the nonlinear distribution of PCC system,a predictive controller is constructed for the operation interval of the system with weak nonlinear capture rate of 50%-90%.A Model Predictive Control（MPC）strategy was designed based on the identification of input and output data in Matlab/Simulink simulation platform.The dynamic and flexible operation of the trap system is improved and the constraint control performance is enhanced.Kalman Filter（KF）was introduced to compensate the adverse effects of the smoke interference on the predictive control effect,which improved the steady-state accuracy and anti-interference ability of the PCC system.（3）To address the strong nonlinearity between multiple variables in the actual PCC process,Nonlinear Model Predictive Control（NMPC）is constructed using open-loop experimental input and output data in the operating condition interval where the PCC system operates with high nonlinear strength.The input and output data are used to identify the PCC system as a Nonlinear Autoregressive Exogenous（NARX）model with exogenous variables,based on which a non-biased nonlinear predictive control strategy is designed and validated by simulation.The results show that the method has excellent control performance and meets the requirements for stable and efficient control of the system.（4）Aiming at the problem that Economic benefit and Control effect are equally important in actual industry,an Economic Model Predictive Control（EMPC）which takes into account economic benefit of PCC system is constructed.A function reflecting the economic performance of the system was established as the objective function of control optimization considering the energy consumption and capture cost of the system,which maximized the economic benefit of the PCC system while maintaining good control performance.