Correlation integral combined with robust predictive control for solving real-time optimization problems and its application
Due to the limitations of traditional correlation integral optimization algorithms in solving real-time optimization problems with no consideration of real-time constraints and difficult to get the gradient adjustment parameter,a real-time optimization algorithm combining robust predictive control and correlation integral is proposed.The traditional correlation integral optimization algorithm is used to calculate the gradients of the optimization objective function and the optimization variables and determine the optimization direction based on the gradients obtained.The optimization gradients are used as the intermediate variables to indicate whether the system still has optimization margin.The constrained robust predictive control based on polytope is used to calculate the adjustments of optimization variables by solving the linear matrix inequalities of the process constraints.The calculated adjustments of optimization variables are used as the set values of the tuning variables.The proposed method inherits the advantages of the original correlation integral optimization algorithm and improves the ability of constraints handling.The simulation and industrial application results for real-time optimization of thermal efficiency of a xylene heater have verified the feasibility and effectiveness of the proposed method.
NETL
NSTL
万方数据
