Robust adaptive model predictive controller with tuning to compensate for model mismatch

• US 8,185,217 B2
• Filed: 01/30/2009
• Issued: 05/22/2012
• Est. Priority Date: 01/31/2008
• Status: Active Grant

• Current Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Original Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Inventors: Thiele, Dirk
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): Sivanesan, Sivalingam
• Application Number:
  US12/363,305
• Publication Number:
  US 20090198350A1
• Time in Patent Office: 1,208 Days
• Field of Search: 700/29, 700/33, 703/2, 703/6, 703/12, 703/13
• US Class Current: 700/29
• CPC Class Codes: G05B 13/042 in which a parameter or coefficient is automatically adjusted to optimise the performance
• CPC Class Codes: G05B 13/048 using a predictor
• CPC Class Codes: G05B 17/02 electric

Abstract

An MPC adaptation and tuning technique integrates feedback control performance better than methods commonly used today in MPC type controllers, resulting in an MPC adaptation/tuning technique that performs better than traditional MPC techniques in the presence of process model mismatch. The MPC controller performance is enhanced by adding a controller adaptation/tuning unit to an MPC controller, which adaptation/tuning unit implements an optimization routine to determine the best or most optimal set of controller design and/or tuning parameters to use within the MPC controller during on-line process control in the presence of a specific amount of model mismatch or a range of model mismatch. The adaptation/tuning unit determines one or more MPC controller tuning and design parameters, including for example, an MPC form, penalty factors for either or both of an MPC controller and an observer and a controller model for use in the MPC controller, based on a previously determined process model and either a known or an expected process model mismatch or process model mismatch range. A closed loop adaptation cycle may be implemented by performing an autocorrelation analysis on the prediction error or the control error to determine when significant process model mismatch exists or to determine an increase or a decrease in process model mismatch over time.

Claims

1. A method of tuning a model predictive controller for use in controlling a process, comprising:

obtaining via a computer processor a process model for the process, the process model including a value for each of a set of process model parameters;

obtaining via a computer processor a process model mismatch indication identifying a process model mismatch for at least one of the set of process model parameters; and

performing on a computer processor a controller optimization based on the process model and the process model mismatch indication including determining a control-based performance measure for the model predictive controller when operated using each of a multiplicity of different sets of controller design/tuning parameter values and the process model in presence of an amount of process model mismatch associated with the process model mismatch indication, and determining an optimal one of the set of the controller design/tuning parameter values for use in the model predictive controller based on the control-based performance measures.

2. The method of claim 1, wherein the control-based performance measure comprises integrated absolute error.

3. The method of claim 1, wherein the controller design/tuning parameters includes a controller form parameter specifying a controller form.

4. The method of claim 3, wherein the controller form parameter indicates one of an observer based model predictive controller form or a non-observer based model predictive controller form.

5. The method of claim 3, wherein the controller form parameter includes an indication of a general Kalman filter observer based model predictive controller form or a simplified Kalman filter observer based model predictive controller form.

6. The method of claim 1, wherein the controller design/tuning parameters includes one or more observer unit tuning parameters.

7. The method of claim 6, wherein the one or more observer unit tuning parameters includes one or more penalty variables.

8. The method of claim 6, wherein the one or more observer unit tuning parameters includes a Kalman filter penalty variable comprising a time constant or a signal to noise ratio or a covariance in a disturbance function or a covariance in a noise function.

9. The method of claim 1, wherein the controller design/tuning parameters includes one or more model predictive control algorithm tuning parameters.

10. The method of claim 9, wherein the one or more model predictive control algorithm tuning parameters includes a penalty variable or a control horizon or a prediction horizon.

11. The method of claim 1, wherein obtaining via a computer processor the process model mismatch indication includes obtaining a specific value of a process model mismatch in one of the process model parameters.

12. The method of claim 1, wherein obtaining via a computer processor the process model mismatch indication includes obtaining specific values of process model mismatch in two or more of the process model parameters.

13. The method of claim 12, wherein the set of process model parameters includes a process gain or a process time constant.

14. The method of claim 1, wherein obtaining via a computer processor the process model mismatch indication includes obtaining a range of process model mismatch in one or more of the process model parameters.

15. The method of claim 1, wherein performing on a computer processor the controller optimization includes determining a control-based performance measure for the model predictive controller when operated using each of a multiplicity of different sets of controller design/tuning parameter values and the process model at each of a plurality of different sets of process model mismatch values to produce a tuning map as a set of points, each point in the tuning map being associated with a determined optimal control-based performance measure obtainable at a particular set of process model mismatch values using the set of controller design/tuning parameter values associated with the determined optimal control-based performance measure.

16. The method of claim 15, including displaying the tuning map on a display device to be visible to a user.

17. The method of claim 16, including allowing a user to choose a particular set of controller design/tuning parameter values to use in tuning the model predictive controller by selecting a particular point in the tuning map.

18. The method of claim 16, wherein obtaining via a computer the process model mismatch indication includes obtaining a model mismatch range for one or more of the set of process model parameters, the model mismatch range defining a subspace in the tuning map, and wherein displaying the tuning map includes displaying the model mismatch range subspace on the tuning map.

19. The method of claim 18, further including allowing a user to move the model mismatch range subspace within the tuning map to determine an optimal adaptation/tuning point in view of the model mismatch range subspace.

20. The method of claim 15, wherein obtaining via a computer processor the process model mismatch indication includes obtaining a model mismatch range, the model mismatch range defining a subspace size in the tuning map and including performing a second optimization to find a particular subspace within the tuning map of the subspace size that produces an optimal value for a second performance measure.

21. The method of claim 20, wherein the second performance measure determines the worst value of the control-based performance measure of the points in the tuning map within the particular subspace.

22. The method of claim 20, including determining a new set of process model parameter values for use in creating a new controller model for the model predictive controller and determining a set of controller design/tuning parameter values to use in the model predictive controller based on the second optimization.

23. The method of claim 15, including determining a new set of controller design/tuning parameter values as the controller design/tuning parameter values associated with one of the points in the tuning map.

24. The method of claim 15, including selecting a new controller operating point as one of the points within the tuning map and providing the controller design/tuning parameter values associated with the selected controller operating point to the model predictive controller for use in controlling the process and providing a new controller model to the model predictive controller based on the process model mismatch values associated with the selected controller operating point.

25. The method of claim 1, wherein obtaining via a computer processor a process model mismatch indication includes performing an autocorrelation procedure on an error signal.

26. The method of claim 25, wherein the error signal is a control error determined in the model predictive controller.

27. The method of claim 26, wherein performing the autocorrelation procedure includes performing the autocorrelation procedure on data collected during a time period when the process is in a steady state condition.

28. The method of claim 25, wherein the error signal is a prediction error determined in the model predictive controller.

29. The method of claim 28, wherein performing the autocorrelation procedure includes performing the autocorrelation procedure on data collected during a time period when the process is being controlled in response to a significant disturbance or upset.

30. An adaptive model predictive controller for use in controlling a process plant, comprising:

a model predictive controller unit having a controller model and one or more alterable design/tuning parameters; and

a tuning unit including;

a model storage that stores a process model for the process plant, the process model specifying a value for each of a set of process model parameters; and

an optimization unit communicatively coupled to the model predictive controller, wherein the optimization unit simulates the operation of the model predictive controller in the presence of a non-zero process model mismatch when the controller model is based on the process model, for each of a plurality of simulation instances, wherein during each simulation instance, the model predictive controller is configured with a different set of controller design/tuning parameter values, the optimization unit determining a controller performance measure for each of the simulation instances, the optimization unit further determining an optimal one of the sets of controller design/tuning parameter values for use in the model predictive controller based on the controller performance measures.

31. The adaptive model predictive controller of claim 30, wherein the controller design/tuning parameters include one or more of a controller form parameter specifying a controller form, or one or more observer unit tuning parameters or one or more model predictive controller algorithm penalty variables.

32. The adaptive model predictive controller of claim 30, wherein the controller design/tuning parameters include a controller form parameter specifying either (1) one of an observer based model predictive controller form or a non-observer based model predictive controller form or (2) one of a first or a second type of observer based model predictive controller form.

33. The adaptive model predictive controller of claim 30, wherein the optimization unit simulates the operation of the model predictive controller in the presence of a plurality of different non-zero process model mismatches to develop a tuning map, the tuning map including a set of points, each point being associated with (1) a determined optimal controller performance measure that is possible at a particular one of the plurality of process model mismatches and (2) the specific set of controller design/tuning parameter values that resulted in the optimal controller performance measure at the particular one of the plurality of process model mismatches.

34. The adaptive model predictive controller of claim 33, further including a display unit communicatively coupled to the optimization unit to display the tuning map to a user.

35. The adaptive model predictive controller of claim 34, wherein the display unit enables a user to select a particular point in the tuning map and wherein the optimization unit provides the design/tuning parameter values associated with the selected point to the model predictive controller.

36. The adaptive model predictive controller of claim 34, wherein the optimization unit stores a model mismatch range for one or more of the process model parameters, the model mismatch range defining a model mismatch range subspace in the tuning map, and wherein the display unit displays the model mismatch range subspace on the tuning map.

37. The adaptive model predictive controller of claim 36, wherein the display unit displays the model mismatch range subspace on the tuning map centered at a point associated with no model mismatch.

38. The adaptive model predictive controller of claim 37, wherein the display unit allows a user to move the model mismatch range subspace within the tuning map to determine an optimal adaptation/tuning point in the tuning map in view of the model mismatch range subspace.

39. The adaptive model predictive controller of claim 33, wherein the optimization unit stores a model mismatch range, the model mismatch range defining a subspace size in the tuning map, and wherein the optimization unit performs a second optimization to determine a particular subspace within the tuning map of the subspace size that produces an optimal value for a second performance measure.

40. The adaptive model predictive controller of claim 39, wherein the optimization unit determines a new set of process model parameter values for use in defining a new controller model for the model predictive controller and determines a set of controller design/tuning parameter values for use in the model predictive controller based on the second optimization.

41. The adaptive model predictive controller of claim 33, wherein the optimization unit determines a new set of controller design/tuning parameter values as the controller design/tuning parameter values associated with one of the points in the tuning map.

42. The adaptive model predictive controller of claim 33, further including an estimator unit that performs an autocorrelation procedure on an error function to implement a tuning cycle for the optimization unit.

43. The adaptive model predictive controller of claim 42, wherein the estimator unit performs an autocorrelation procedure on the error function to estimate an amount of process model mismatch in one of the process model parameters.

44. The adaptive model predictive controller of claim 42, wherein the estimator unit performs an autocorrelation procedure on a control error determined in the model predictive controller.

45. The adaptive model predictive controller of claim 42, wherein the estimator unit performs an autocorrelation procedure a control error determined in the model predictive controller during a steady state operation of the process.

46. An adaptive model predictive controller tuning unit for implementation on a computer processor to tune a model predictive controller that operates to control a process plant using a controller model, comprising:

a non-transitory computer readable medium;

a storage routine stored on the computer readable medium for execution on a processor to store a process model for the process plant, the process model specifying a value for each of a set of process model parameters; and

an optimization routine stored on the computer readable medium for execution on a processor to simulate the operation of the model predictive controller in presence of a process model mismatch when the controller model of the model predictive controller is based on the process model, the optimization routine simulating the model predictive controller when the model predictive controller is configured with each of a plurality of different sets of controller design/tuning parameter values, the optimization routine determining a controller performance measure for each of the plurality of different sets of controller design/tuning parameter values in the presence of the process model mismatch, and the optimization routine still further determining an optimal one of the different sets of controller design/tuning parameter values for use in the model predictive controller based on the controller performance measures.

47. The adaptive model predictive controller tuning unit of claim 46, wherein the controller design/tuning parameters include a controller form parameter specifying a controller form, or one or more model predictive controller observer unit tuning parameters, or one or more model predictive controller algorithm penalty variables.

48. The adaptive model predictive controller tuning unit of claim 46, wherein the optimization routine executes to simulate the operation of the model predictive controller in the presence of a plurality of different process model mismatches to develop a tuning map, the tuning map including a set of points, each point being associated with (1) a determined optimal controller performance measure that is possible at a particular one of the plurality of process model mismatches and (2) the specific set of controller design/tuning parameter values that resulted in the optimal controller performance measure at the particular one of the plurality of process model mismatches.

49. The adaptive model predictive controller tuning unit of claim 48, further including a display routine communicatively coupled to the optimization routine to display the tuning map to a user, and wherein the display routine enables a user to select a particular point in the tuning map to provide the design tuning parameter values associated with the selected point to the model predictive controller for tuning the model predictive controller.

50. The adaptive model predictive controller tuning unit of claim 49, wherein the optimization routine includes a model mismatch range for one or more of the process model parameters, the model mismatch range defining a subspace in the tuning map, and wherein the display unit displays the model mismatch range subspace on the tuning map.

51. The adaptive model predictive controller tuning unit of claim 48, wherein the optimization routine executes to perform a second optimization to determine a particular model mismatch subspace within the tuning map that produces an optimal value for a second performance measure wherein the model mismatch subspace defines a model mismatch range in one or more of the process model parameters.

52. The adaptive model predictive controller tuning unit of claim 51, wherein the optimization routine executes to determine a new set of process model parameter values for use in defining a new controller model for the model predictive controller and determines a set of controller design/tuning parameter values for use in the model predictive controller based on the second optimization.