Margin saving model predictive control with operating point back-off

作者：Luo Xionglin^*; Zhou Yantong; Xu Feng

来源：International Journal of Innovative Computing Information and Control, 2018, 14(2): 405-421.

• Conventional modeling
• Manipulated variables
• Margin loss functions
• Margin saving
• Multivariable control
• Operation optimization
• Process system
• Steady-state analysis

In chemical processes, design margins must be added on the nominal values of design variables during process design period, so the operating point will move into the feasible domain and there are sufficient operating margins between current operating point and process constraints. When plants running, process operators and engineers usually hope the operating margin consumption to be as little as possible in dynamic control process in order to keep more operating space for operation optimization. Model predictive control can deal with the problems of constrained control and multivariable control, but the optimal control sequence may reach the boundary of the feasible domain due to constraints. Therefore, a margin saving model predictive control algorithm is proposed through adding a margin loss function indicating operating margin reduction amount to the optimal control performance index function of conventional model predictive control, making the operating point back-off from the boundary of the feasible domain. Through the steady-state analysis, the margin saving effects for manipulated variables and the tracking effects for controlled variables can be found. Finally, a simulation case is given to illustrate the effectiveness of the algorithm.

• 出版日期2018
• 单位中国石油大学（北京）