Method and apparatus for attenuating error in dynamic and steady-state processes for prediction, control, and optimization

• US 20060229743A1
• Filed: 02/21/2006
• Published: 10/12/2006
• Est. Priority Date: 05/06/1996
• Status: Active Grant

• Current Assignee: Rockwell Automation Technologies Incorporated (Allen-Bradley Co., Inc.)
• Original Assignee: Rockwell Automation, Inc.
• Inventors: Piche, Stephen, Martin, Gregory D., Boe, Eugene
• Granted Patent: US 7,610,108 B2
• Time in Patent Office: Days
• US Class Current: 700/44
• CPC Class Codes: G05B 13/027 using neural networks only
• 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

Abstract

A method for providing independent static and dynamic models in a 000 prediction, control and optimization environment utilizes an independent static model (20) and an independent dynamic model (22). The static model (20) is a rigorous predictive model that is trained over a wide range of data, whereas the dynamic model (22) is trained over a narrow range of data. The gain K of the static model (20) is utilized to scale the gain k of the dynamic model (22). The forced dynamic portion of the model (22) referred to as the b_i variables are scaled by the ratio of the gains K and k. The b_i have a direct effect on the gain of a dynamic model (22). This is facilitated by a coefficient modification block (40). Thereafter, the difference between the new value input to the static model (20) and the prior steady-state value is utilized as an input to the dynamic model (22). The predicted dynamic output is then summed with the previous steady-state value to provide a predicted value Y. Additionally, the path that is traversed between steady-state value changes.

Claims

1. A method for predicting an output value from a received input value, the method comprising:

modeling a set of static data received from a plant or process in a predictive static model over a first range, the static model having a static gain K and modeling static operation of the plant or process;

modeling a set of dynamic data received from the plant or process in a predictive dynamic model over a second range smaller than the first range, the dynamic model having a dynamic gain k and modeling dynamic operation of the plant or process over the second range, and the dynamic model being independent of operation of the static model;

adjusting the dynamic gain k of the dynamic model as a predetermined function of the static gain K of the static model to vary model parameters of the dynamic model;

predicting the dynamic operation of the plant or process from an initial steady-state input value at a first time to a predicted steady-state input value at a second time to determine a predicted dynamic operation;

comparing the predicted dynamic operation to a desired steady-state value of the plant or process at a final desired output value and generating an error as the difference therebetween;

attenuating the error as a function of time between the first time and the second time;

determining a change in the input value for input to said predicting the dynamic operation which is operable to vary the input value thereto; and

varying the change in the input value in accordance with the determined change to minimize the attenuated error.

2. The method of claim 1, wherein said attenuating the error comprises:

attenuating the error as a function of the values of the gain k at the first time and the second time.

3. The method of claim 2, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error as a function of a ratio of the values of the gain k at the first time and the second time.

4. The method of claim 2, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a simple average of the values of the gain k at the first time and the second time.

5. The method of claim 2, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a larger absolute magnitude value of the values of the gain k at the first time and the second time.

6. The method of claim 2, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a weighted average of the values of the gain k at the first time and the second time.

7. The method of claim 2, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to one or more values between the values of the gain k at the first time and the second time inclusively.

8. The method of claim 2, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a linear interpolation between the values of the gain k at the first time and the second time inclusively.

9. The method of claim 2, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a nonlinear interpolation between the values of the gain k at the first time and the second time inclusively.

10. The method of claim 1, further comprising:

utilizing one or more of the static model and the dynamic model to control the plant or process.

11. A computer-accessible memory medium for predicting an output value from a received input value, the memory medium comprising computer instructions executable by a processor to perform:

modeling a set of static data received from a plant or process in a predictive static model over a first range, the static model having a static gain K and modeling static operation of the plant or process;

modeling a set of dynamic data received from the plant or process in a predictive dynamic model over a second range smaller than the first range, the dynamic model having a dynamic gain k and modeling dynamic operation of the plant or process over the second range, and the dynamic model being independent of operation of the static model;

adjusting the dynamic gain k of the dynamic model as a predetermined function of the static gain K of the static model to vary model parameters of the dynamic model;

predicting the dynamic operation of the plant or process from an initial steady-state input value at a first time to a predicted steady-state input value at a second time to determine a predicted dynamic operation;

comparing the predicted dynamic operation to a desired steady-state value ofthe plant or process at a final desired output value and generating an error as the difference therebetween;

attenuating the error as a function of time between the first time and the second time;

determining a change in the input value for input to said predicting the dynamic operation which is operable to vary the input value thereto; and

varying the change in the input value in accordance with the determined change to minimize the attenuated error.

12. The memory medium of claim 11, wherein said attenuating the error comprises:

attenuating the error as a function of the values of the gain k at the first time and the second time.

13. The memory medium of claim 12, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error as a function of a ratio of the values of the gain k at the first time and the second time.

14. The memory medium of claim 12, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a simple average of the values of the gain k at the first time and the second time.

15. The memory medium of claim 12, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a larger absolute magnitude value of the values of the gain k at the first time and the second time.

16. The memory medium of claim 12, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a weighted average of the values of the gain k at the first time and the second time.

17. The memory medium of claim 12, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to one or more values between the values of the gain k at the first time and the second time inclusively.

18. The memory medium of claim 12, wherein said attenuating the error as a function of the values of the gain k at the first time and the second time comprises:

attenuating the error according to a linear interpolation between the values of the gain k at the first time and the second time inclusively.

19. A system for predicting an output value from a received input value, comprising:

a predictive static model, wherein the predictive static models has a static gain K and models static operation of the plant or process, and wherein the predictive static model comprises a set of static data over a first range received from a plant or process;

a dynamic model coupled to the predictive static model, wherein the dynamic model has a dynamic gain k and models dynamic operation of the plant or process, and wherein the dynamic model comprises a set of dynamic data over a second range smaller than the first range received from the plant or process;

wherein the dynamic model is operable to;

adjust the dynamic gain k as a predetermined function of the static gain K of the static model to vary model parameters of the dynamic model;

predict the dynamic operation of the plant or process from an initial steady-state input value at a first time to a predicted steady-state input value at a second time to determine a predicted dynamic operation;

compare the predicted dynamic operation to a desired steady-state value of the plant or process at a final desired output value and generate an error as the difference therebetween;

attenuate the error as a function of time between the first time and the second time;

determine a change in the input value for input to the dynamic model which is operable to vary the input value thereto; and

vary the change in the input value in accordance with the determined change to minimize the attenuated error.

20. A system, comprising:

means for modeling a set of static data received from a plant or process in a predictive static model over a first range, the static model having a static gain K and modeling static operation of the plant or process;

means for modeling a set of dynamic data received from the plant or process in a predictive dynamic model over a second range smaller than the first range, the dynamic model having a dynamic gain k and modeling dynamic operation of the plant or process over the second range, and the dynamic model being independent of operation of the static model;

means for adjusting the dynamic gain k of the dynamic model as a predetermined function ofthe static gain K ofthe static model to vary model parameters of the dynamic model;

means for predicting the dynamic operation of the plant or process from an initial steady-state input value at a first time to a predicted steady-state input value at a second time to determine a predicted dynamic operation;

means for comparing the predicted dynamic operation to a desired steady-state value of the plant or process at a final desired output value and generating an error as the difference therebetween;

means for attenuating the error as a function of time between the first time and the second time;

means for determining a change in the input value for input to said predicting the dynamic operation which is operable to vary the input value thereto; and

means for varying the change in the input value in accordance with the determined change to minimize the attenuated error.