MULTI-OBJECTIVE PREDICTIVE PROCESS OPTIMIZATION WITH CONCURRENT PROCESS SIMULATION

• US 20070168057A1
• Filed: 12/05/2006
• Published: 07/19/2007
• Est. Priority Date: 12/05/2005
• Status: Active Grant

• Current Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Original Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Inventors: Nixon, Mark, Wojsznis, Peter, Blevins, Terrence, Wojsznis, Wilhelm
• Granted Patent: US 8,055,358 B2
• Time in Patent Office: Days
• US Class Current: 700/53
• CPC Class Codes: G05B 13/022 using a perturbation of the variable
• CPC Class Codes: G05B 13/048 using a predictor
• CPC Class Codes: G05B 17/02 electric

Abstract

A system and method for controlling a process includes simulating the process and producing a simulated output of the process, developing a set of target values based on measured inputs from the process and based on the simulated output from the process simulator, and producing multiple control outputs configured to control the process based on the set of target values during each operational cycle of the process control system. The simulated outputs include one or more predicted future values up to the steady state of the process.

Claims

1. A process control system for controlling a process, comprising:

a process simulator adapted to simulate the process and produce a simulated output of the process, wherein the simulated output includes one or more predicted future values of the process;

an optimizer adapted to develop a set of target values based on measured inputs from the process and based on the simulated output from the process simulator; and

a multiple-input/multiple-output controller adapted to produce, during each operational cycle of the process control system, multiple control outputs configured to control the process based on the set of target values provided to the multiple-input/multiple output controller from the optimizer during each operational cycle of the process control system.

2. The process control system of claim 1, wherein the process simulator is adapted to simulate the process concurrently with the operation of the process.

3. The process control system of claim 1, wherein the process simulator is adapted to simulate the process based on measured feedback outputs from the process, wherein simulation results are corrected by the measured feedback outputs to produce the simulated output of the process.

4. The process control system of claim 1, wherein the process simulator is adapted to simulate the process independent of the optimizer.

5. The process control system of claim 1, wherein the process simulator comprises a simulation model of the process and the optimizer comprises an optimization steady state model of the process based on the simulation model.

6. The process control system of claim 1, wherein the process simulator is adapted to simulate the process and produce a simulated output of the process up to a steady state of the process, and wherein the optimizer is adapted to develop the set of target values based on multiple measured inputs from the process and based on the simulated output of the process from the process simulator up to the steady state of the process.

7. The process control system of claim 1, wherein the process simulator comprises a step response model of the process.

8. The process control system of claim 7, wherein the process simulator is adapted to utilize a dynamic model matrix determined from the step response model to simulate the process and produce a simulated output of the process.

9. The process control system of claim 1, wherein the process simulator is adapted to utilize a dynamic model matrix determined from at least one of the group consisting of:

first principles modeling and parametric modeling.

10. The process control system of claim 9, wherein the optimizer is adapted to develop a set of target values based upon a steady state gain matrix, wherein the steady state gain matrix is based upon the dynamic model matrix, and wherein the process simulator is adapted to update the dynamic model matrix in real time.

11. The process control system of claim 1, wherein the process simulator comprises one or more simulation blocks configured to simulate one or more portions of the process, each simulation block comprising:

a process model that models at least a portion of the process; and

a simulation unit that utilizes the process model to simulate the operation of the portion of the process to produce the simulated output for a process element associated with the portion of the process, the simulated output including or more predicted future values for the process element.

12. The process control system of claim 11, wherein at least one of the simulation blocks includes an input adapted to accept an actual value of the process element corresponding to the simulated output for the process element.

13. The process control system of claim 12, wherein the input is communicatively connected to accept a measurement of the process element made during on-line operation of the process.

14. The process control system of claim 1, wherein the optimizer is adapted to apply the target values to the multiple-input/multiple-output controller as target manipulated variables.

15. The process control system of claim 1, wherein the optimizer is adapted to apply the target values to the multiple-input/multiple-output controller as target control variables.

16. The process control system of claim 1, wherein the optimizer is adapted to develop the set of target values based on an objective function and based on flexible process limitations.

17. The process control system of claim 1, wherein the optimizer comprises an objective function, and the optimizer is adapted to minimize or maximize the objective function and is further adapted to extend the objective function by degrees of freedom beyond a predefined set point range of a control variable.

18. The process control system of claim 1, wherein the optimizer comprises an objective function and the optimizer is adapted to minimize or maximize the objective function while keeping a set of control variables within predefined control variable limits and a set of manipulated variables within a set of predefined manipulated variable limits and, if no solution exists, to enable at least one of the control variable limits to be violated.

19. The process control system of claim 18, wherein the control variables comprise predicted control variables determined from the simulated output of the process.

20. The process control system of claim 18, wherein the optimizer is adapted to utilize one or more slack variables in the objective function to extend the set of control variables beyond the control variable limits, and further adapted to develop the set of target values based on a solution determined outside the predefined control variable limits.

21. The process control system of claim 20, wherein the optimizer is adapted to incrementally include slack variables into the objective function, wherein each additional slack variable extends the objective function by a further degree of freedom beyond the predefined set point limits.

22. The process control system of claim 21, wherein the optimizer is adapted to apply a penalty to each additional slack variable included in the objective function, wherein the penalty for each slack variable is dependent on a priority associated with the control variable.

23. The process control system of claim 1, wherein the optimizer comprises a linear programming optimizer.

24. The process control system of claim 1, wherein the optimizer comprises a nonlinear programming optimizer.

25. The process control system of claim 1, wherein the multiple-input/multiple-output controller comprises a model predictive controller.

26. A method of optimizing a process having a plurality of manipulated variables and a plurality of control variables capable of being effected by changes in the manipulated variables, the method comprising:

simulating the operation of the process using a process model of the process to produce a simulated output of the process, wherein the simulated output includes one or more predicted future values of the process;

providing the predicted future values of the process to an optimization model of the portion of the process;

measuring inputs from the process;

performing process optimization using an optimization model by selecting a process operating point based on the measured inputs from the process and based on the simulated output from the process simulator, wherein the process operating point is defined by a set of target values for one or more of the plurality of control variables; and

using the set of target values to control the process.

27. The method of claim 26, wherein simulating the operation of the process comprises simulating the operation of the process concurrently with the operation of the process.

28. The method of claim 26, wherein simulating the operation of the process comprises:

measuring feedback outputs from the process, and

correcting simulation results based on the measured feedback outputs from the process.

29. The method of claim 26, wherein performing process optimization comprises performing process optimization independent of simulating the operation of the process.

30. The method of claim 26, wherein simulating the operation of the process comprises simulating the operation of the process at a steady state of the process to produce a simulated output of the steady state of the process.

31. The method of claim 26, wherein performing process optimization comprises minimizing or maximizing an objective function with flexible process limitations.

32. The method of claim 26, wherein simulating the operation of the process using a process model of the process to produce a simulated output of the process, comprising predicting the one or more control variables.

33. The method of claim 31, wherein minimizing or maximizing an objective function with flexible process limitations comprises extending the objective function by degrees of freedom beyond a predefined set point range of a control variable.

34. The method of claim 31, wherein minimizing or maximizing an objective function with flexible process limitations comprises:

maintaining the one or more control variables within predefined control variable limits;

maintaining one or more of the manipulated variables within a set of predefined manipulated variable limits;

violating at least one of the control variable limits if no solution exists within the predefined limits; and

selecting the process operating point based on a solution determined outside of the predefined limits.

35. The method of claim 34, wherein violating at least one of the control variable limits comprises applying one or more slack variables to the objective function, wherein each slack variable extends the one or more control variables beyond the predefined set point limits.

36. The method of claim 35, further comprising applying a penalty to each slack variable included in the objective function.

37. A process control element adapted to be used as a portion of a process control routine implemented on a processor to control a plurality of control parameters of a process using a plurality of manipulated parameters, the process control element comprising:

a computer readable medium;

a simulation function block stored on the computer readable medium and adapted to be executed on the processor to implement simulation of the process, the simulation function block comprising a simulation routine that simulates the process and produces a simulated output of the process including one or more predicted future values of the process;

a optimization function block stored on the computer readable medium and adapted to be executed on the processor to implement multiple input/multiple output control of the process during, the optimization function block including;

an objective function that defines an optimization criteria based on the plurality of control parameters;

an optimizer routine that uses the objective function and the simulated output from the simulation routine to produce a set of optimal target values for the control parameters during each control scan period;

a multiple-input/multiple-output control routine that produces a control signal for each of the plurality of manipulated parameters during each control scan period using the target values.

38. The process control element of claim 37, wherein the simulation routine simulates the process concurrently with the operation of the process.

39. The process control element of claim 37, wherein the simulation routine utilizes measured feedback outputs from the process to correct the simulation of the process to produce the simulated output of the process.

40. The process control element of claim 37, wherein the optimization function block comprises a steady state model of the process.

41. The process control element of claim 40, wherein the simulation function block comprises a simulation model of the process, and wherein the steady state model of the process is based on the simulation model of the process.

42. The process control element of claim 37, wherein the simulation routine simulates the process and produces a simulated output of the process to a steady state of the process, and wherein the optimizer routine uses the objective function and the simulated steady state output of the process to produce the set of optimal target values for the control parameters.

43. The process control element of claim 37, wherein the simulation function block comprises a model of the process and wherein the simulation routine utilizes a dynamic model matrix determined from the model.

44. The process control element of claim 43, wherein the model comprises at least one of the group consisting of:

a step response model of the process, a first principle model of the process and a parametric model of the process.

45. The process control element of claim 43, wherein optimization function block comprises a steady state gain matrix based upon the dynamic model matrix, wherein the optimizer routine uses the steady state gain matrix to produce the set of optimal target values, and wherein the simulation routine updates the dynamic model matrix in real time.

46. The process control element of claim 37, wherein an optimizer routine applies the target values to the multiple-input/multiple-output control routine as target manipulated variables.

47. The process control element of claim 37, wherein an optimizer routine applies the target values to the multiple-input/multiple-output control routine as target control variables.

48. The process control element of claim 37, wherein the objective function comprises at least one slack variable and predefined control variable limits, and wherein the optimizer routine minimizes or maximizes the objective function while keeping the control variables within the predefined control variable limits, and, if an optimal solution does not exist within the predefined control variable limits, the optimizer routine violates the predefined control variable limits based on the at least one slack variable to produce the set of optimal target values for the control parameters.

49. The process control element of claim 48, wherein each slack variable comprises a degree of freedom beyond the control variable limits.

50. The process control element of claim 48, wherein the optimizer routine incrementally applies each slack variable to the objective function if an optimal solution does not exist within the predefined control variable limits and within any previously applied slack variables.

51. The process control element of claim 50, wherein the optimizer routine applies a penalty to each incrementally-applied slack variable.

52. The process control element of claim 37, wherein the optimizer routine comprises at least one of the group consisting of:

a linear optimizer routine and a nonlinear optimization routine.