Dynamic on-line optimization of production processes

A production process system, comprising:

• a number of process sections, controllable by manipulated variables;
• sensors measuring a set of process output variables of said production process;
• processor means, connected to said sensors;
• process section control means, setting said manipulated variables, said process section control means being connected to said processor means;
• said processor means in turn comprising:
• means for obtaining a dynamic process model having time dependent relations between manipulated variables for said process sections and process output variables from respective process section;
• means for providing external constraints for a prediction time interval;
• means for defining an objective function involving predicted controlled process output variables for said prediction time interval and said external constraints;
• said predicted controlled process output variables being defined by said dynamic process model based on an estimation of an initial state obtained by said measured set of process output variables;
• means for optimizing said objective function under constraints imposed by said dynamic process model and/or said external constraints by adapting said manipulated variables, giving input trajectories for said manipulated variables for said prediction time interval;
• whereby said process section control means is arranged for setting said manipulated variables according to said input trajectories during a control time interval.

A system for a production process having a number of process sections, controllable by manipulated variables, comprising:

• processor means;
• and
• process section control means, setting said manipulated variables, said process section control means being connected to said processor means;
• said processor means in turn comprising:
• sensor input means for receiving a set of process output variables of said production process;
• means for obtaining a dynamic process model having time dependent relations between manipulated variables for said process sections and process output variables from respective process section;
• means for providing external constraints for a prediction time interval;
• means for defining an objective function involving predicted controlled process output variables for said prediction time interval and said external constraints;
• said predicted controlled process output variables being defined by said dynamic process model based on an estimation of an initial state obtained by said measured set of process output variables;
• means for optimizing said objective function under constraints imposed by said dynamic process model and/or said external constraints by adapting said manipulated variables, giving input trajectories for said manipulated variables for said prediction time interval;
• whereby said process section control means is arranged for setting said manipulated variables according to said input trajectories during a control time interval.