Dynamic Controller Utilizing a Hybrid Model

• US 20080065241A1
• Filed: 09/13/2006
• Published: 03/13/2008
• Est. Priority Date: 09/13/2006
• Status: Active Grant

• Current Assignee: Rockwell Automation Technologies Incorporated (Allen-Bradley Co., Inc.)
• Original Assignee: Rockwell Automation, Inc.
• Inventors: Martin, Gregory D., Boe, Eugene, Piche, Stephen W.
• Granted Patent: US 7,496,414 B2
• Time in Patent Office: Days
• US Class Current: 700/44
• 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

A system and method for predicting operation of a plant or process receive an input value from the plant or process. An integrity of a non-linear model corresponding to a local input space of the input value may be determined. The non-linear model may include an empirical representation of the plant or process. If the integrity is above a first threshold, non-linear model may be used to provide a first output value. However, if the integrity is below the first threshold, a linearized first principles model may be used to provide a second output value. The linearized first principles model may include an analytic representation of the plant or process. Additionally, the analytic representation of the plant or process may be independent of the empirical representation of the plant or process. The first output value and/or the second output value may be usable to manage the plant or process.

Claims

1. A predictive system for predicting operation of a plant or process, comprising:

a non-linear model comprising;

an empirical representation of the plant or process;

a first input operable to receive an input value; and

a first output operable to provide a first output value;

wherein the non-linear model is operable to receive the input value and generate the first output value using the empirical representation of the plant or process;

a linearized first principles model comprising;

an analytic representation of the plant or process, wherein the analytic representation of the plant or process is independent of the empirical representation;

a second input operable to receive the input value; and

a second output operable to provide a second output value;

wherein the linearized first principles model is operable to receive the input value and generate the second output value using the analytic representation of the plant or process;

a domain analyzer operable to determine whether the input value falls within a region of space where the non-linear model has an integrity that is less than a first threshold; and

a controller operable to;

use the non-linear model when the integrity is above the first threshold; and

use the linearized first principles model when the integrity is below the first threshold;

wherein the first output value and/or the second output value is usable to manage the plant or process.

2. The system of claim 1, wherein the linearized first principles model comprises a linearization of a first principles model.

3. The system of claim 1, wherein the linearized first principles model comprises a local linearization of a first principles model local to the input value.

4. The system of claim 1, wherein the linearized first principles model comprises a global linearization of a first principles model.

5. The system of claim 4, wherein the global linearization is based on the range of previous input values.

6. The system of claim 1, wherein the domain analyzer is further operable to determine whether the input value falls within a region of local input space where the non-linear model has an integrity that is less than a second threshold, and wherein the second threshold is lower than the first threshold.

wherein the controller is operable to:

use the non-linear model when the integrity is above the first threshold;

use a first principles model when the integrity is above the second threshold and below the first threshold; and

use the linearized first principles model when the integrity is below the second threshold.

7. The system of claim 1, wherein the non-linear model comprises a steady-state model.

8. The system of claim 1, wherein the non-linear model comprises:

a neural network;

or

a support vector machine.

9. A method for predicting operation of a plant or process, comprising:

receiving an input value from the plant or process;

determining an integrity of a non-linear model corresponding to a local input space of the input value, wherein the non-linear model comprises an empirical representation of the plant or process;

if the integrity is above a first threshold, using the non-linear model to provide a first output value; and

if the integrity is below the first threshold, using a linearized first principles model to provide a second output value, wherein the linearized first principles model comprises an analytic representation of the plant or process, and wherein the analytic representation of the plant or process is independent of the empirical representation of the plant or process;

wherein the first output value and/or the second output value is usable to manage the plant or process.

10. The method of claim 9, wherein the linearized first principles model comprises a linearization of a first principles model.

11. The method of claim 9, wherein the linearized first principles model comprises a local linearization of a first principles model local to the input value.

12. The method of claim 9, wherein the linearized first principles model comprises a global linearization of a first principles model.

13. The method of claim 12, wherein the global linearization is based on the range of previous input values.

14. The method of claim 9, wherein the method further comprises:

using the non-linear model when the integrity is above the first threshold and a second threshold, wherein the second threshold is lower than the first threshold;

using a first principles model when the integrity is above the second threshold; and

using the linearized first principles model when the integrity is below the second threshold.

15. The method of claim 9, wherein the non-linear model comprises a steady-state model.

16. A memory medium comprising program instructions for predicting operation of a plant or process, wherein the program instructions are executable by a processor to:

receive an input value from the plant or process;

determine an integrity of a non-linear model corresponding to a local input space of the input value, wherein the non-linear model comprises an empirical representation of the plant or process;

if the integrity is above a first threshold, use the non-linear model to provide a first output value; and

if the integrity is below the first threshold, use a linearized first principles model to provide a second output value, wherein the linearized first principles model comprises an analytic representation of the plant or process, and wherein the analytic representation of the plant or process is independent of the empirical representation of the plant or process;

wherein the first output value and/or the second output value is usable to manage the plant or process.

17. The memory medium of claim 16, wherein the linearized first principles model comprises a linearization of a first principles model.

18. The memory medium of claim 16, wherein the linearized first principles model comprises a local linearization of a first principles model local to the input value.

19. The memory medium of claim 16, wherein the linearized first principles model comprises a global linearization of a first principles model.

20. The memory medium of claim 16, wherein the program instructions are further executable to:

use the non-linear model when the integrity is above the first threshold and a second threshold, wherein the second threshold is lower than the first threshold;

use a first principles model when the integrity is above the second threshold; and

use the linearized first principles model when the integrity is below the second threshold.