System and method for superior performance with respect to best performance values in model predictive control applications
Model predictive control is used to obtain the best performance value for an objective in a dynamic environment. One or more best performance values for a model predictive application are obtained by utilizing asymmetric dynamic behavior that pushes the process to the edges of the operative window. When a setpoint becomes infeasible, a controller slows down when moving away from a specified setpoint. When the infeasibility clears and the controller starts moving back towards the setpoint, the controller follows a tuned speed. When the controller moves in a direction against economic profit, the controller slows down and when moving in the direction of economic profit, the controller follows the tuned dynamic speed. The controller computes the best performance value for each controlled variable and is equal to the setpoint for the controlled variables with a setpoint or the highest profit value between limits for controlled values affected by economic functions.
1. A method for high dynamic performance control of control variables in a model predictive controller, comprising the steps of:
specifying a setpoint for a controlled variable;
specifying economic functions driving a controller variable;
specifying relative prioritization between control objectives;
determining a steady state target value;
determining one or more best performance values;
determining a stage cost;
detecting a disturbance;
determining one or more dynamic tuning weights;
adjusting the one or more best performance values;
creating dynamic control objective using the adjusted one or more best performance values;
computing dynamic plan of actions for manipulated variables; and
executing the dynamic plan of actions for manipulated variables with respect to an environment wherein the dynamic plan of actions exhibits asymmetric behavior such that whenever a setpoint becomes infeasible, the controller slows down when moving away from the setpoint and when the setpoint becomes feasible the controller follows the tuned speed when moving towards the setpoint.
2. The method of claim 1 , wherein the setpoint of the controlled variable comprises an upper setpoint and a lower setpoint.
3. The method of claim 1 , further comprising:
specifying a second setpoint for a second controlled variable, specifying at least one of economic objectives, and relative prioritization of control objectives and economic function.
4. The method of claim 1 , wherein the adjusted best performance values are based, at least in part, on the detected disturbance.
5. The method of claim 1 , wherein the detected disturbance comprises one or more of change in temperature, quality, control, level, and volume.
6. The method of claim 1 , wherein the steady state target value is based, at least in part, on a specified static model.
7. The method of claim 1 , wherein the stage cost is based, at least in part, on one or more assessed penalties.
8. A system for control of controlled variables in a model predictive controller, the system comprising:
one or more processors for processing information;
a memory communicatively coupled to the one or more processors; and
one or more controllers that comprise instructions stored in the memory, the instructions, when executed by the processor, operable to perform operations comprising:
specifying a setpoint for a controlled variable;
determining a steady state target value;
determining one or more best performance values;
determining a stage cost;
detecting a disturbance;
determining one or more dynamic tuning weights;
adjusting the one or more best performance values;
creating dynamic control objective using the adjusted one or more best performance values;
adjusting a control model; and
executing the adjusted control model with respect to an environment wherein the dynamic plan of actions exhibits asymmetric behavior such that whenever a setpoint becomes infeasible, the controller slows down when moving away from the setpoint and when the setpoint becomes feasible the controller follows the tuned speed when moving towards the setpoint.
9. The system of claim 8 , wherein the setpoint of the controlled variable comprises an upper setpoint and a lower setpoint.
10. The system of claim 8 , further comprising:
specifying a second setpoint for a second controlled variable.
11. The system of claim 8 , wherein the adjusted best performance values are based, at least in part, on the detected disturbance.
12. The system of claim 8 , wherein the detected disturbance comprises one or more of change in temperature, quality, control, level, and volume.
13. The system of claim 8 , wherein the steady state target value is based, at least in part, on a specified static model.
14. The system of claim 8 , wherein the stage cost is based, at least in part, on one or more assessed penalties.
15. A method for high dynamic performance control of control variables in a model predictive controller, comprising the steps of:
specifying a setpoint for a controlled variable;
specifying economic functions driving a controller variable;
specifying relative prioritization between control objectives;
determining a steady state target value;
determining one or more best performance values wherein the best performance value is the value that corresponds to the best performance within specified limits and is equal to the setpoint for controlled variables or the highest profit value between limits for controlled variables affected by economic functions;
determining a stage cost;
detecting a disturbance;
determining one or more dynamic tuning weights;
adjusting the one or more best performance values;
creating dynamic control objective using the adjusted one or more best performance values;
computing dynamic plan of actions for manipulated variables; and
executing the dynamic plan of actions for manipulated variables with respect to an environment wherein the dynamic plan of actions exhibits asymmetric behavior such that whenever a setpoint becomes infeasible, the controller slows down when moving away from the setpoint and when the setpoint becomes feasible the controller follows the tuned speed when moving towards the setpoint.