Kiln thermal and combustion control

• US 6,438,430 B1
• Filed: 05/09/2000
• Issued: 08/20/2002
• Est. Priority Date: 05/06/1996
• Status: Expired due to Term

• Current Assignee: Rockwell Automation Technologies Incorporated (Rockwell Automation, Inc.)
• Original Assignee: Pavilion Technologies Incorporated (Rockwell Automation, Inc.)
• Inventors: Piche, Stephen, Martin, Gregory D., Boe, Eugene, Havener, John P., Gerules, Mark, Timmer, Douglas, Keeler, James David
• Primary Examiner(s): Picard, Leo
• Assistant Examiner(s): Bahta, Kidest
• Application Number:
  US09/568,190
• Time in Patent Office: 833 Days
• Field of Search: 700/23, 700/29, 700/30, 700/34, 700/40-44, 700/52, 219/497, 703/12
• US Class Current: 700/28
• 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 kiln thermal and combustion control. A predictive model is provided of the dynamics of selected aspects of the operation of the plant for modeling the dynamics thereof. The model has at least two discrete models associated therewith that model at least two of the selected aspects, the at least two discrete models having different dynamic responses. An optimizer receives desired values for the selected aspects of the operation of the plant modeled by the model and optimizes the inputs to the model to minimize error between the predicted and desired values. A control input device then applies the optimized input values to the plant after optimization thereof.

Claims

1. A controller for controlling a multi-variable input plant having a plurality of manipulatible variables (MVs) as inputs, and operable to provide an output, comprising:

a predictive model of the dynamics of selected aspects of the operation of the plant for modeling the dynamics thereof, said model having at least two discrete models associated therewith that model at least two of the selected aspects, said at least two discrete models having different dynamic responses and each of said discrete models providing a predicted output of said respective selected aspect;

an optimizer for receiving desired values for the selected aspects of the operation of the plant modeled by said predictive model and said predicted outputs from said predictive model and optimizing the inputs to said predictive model to minimize error between the predicted and desired values; and

a control input device for applying the optimized input values to the plant after optimization thereof.

2. The controller of claim 1, wherein said at least two discrete models are linear dynamic models.

3. The controller of claim 1, wherein said at least two discrete models are independent of each other.

4. The controller of claim 1, wherein said at least two discrete models are separate and independent models.

5. The controller of claim 1, wherein said optimizer is further operable to receive constraints on said input values such that the optimization operation changes the inputs within said constraints when minimizing the error between the predicted and desired values.

6. The controller of claim 1, wherein the plant is subject to at least one disturbance variable that alters the operation of the plant and wherein said predictive model will predict the behavior of the selected aspects modeled thereby in response to changes in the at least one disturbance variable and said optimizer will change the inputs to again minimize the error between the predicted and desired values.

7. The controller of claim 1, wherein said two discrete models are linear dynamic models and wherein each said at least two discrete models each comprise:

a plurality of independent sub models, each associated with one of the inputs to the associated model, each of said independent sub models operable to model the associated select aspect as a function of the associated input thereto; and

a summing device for summing the output of said independent sub models for each of said discrete models to provide a single output associated with each of said discrete models that is a function of all of the inputs that are input to each of said associated sub models for each of said discrete models.

8. The controller of claim 7, wherein said optimizer comprises:

a comparator for comparing for each of said discrete models the associated desired value with the predicted output of said discrete model to generate an error value as the difference therebetween for each of said discrete models; and

an iterative optimization device for incrementally changing the MV inputs over a defined dynamic path for all of said discrete models and inputting them to said model over a plurality of iterative cycles to minimize the error between the desired values and the predicted values.

9. The controller of claim 8, wherein said iterative controller optimization device is operable to utilize a steepest descent optimization algorithm for the optimization procedure.

10. The controller of claim 1, wherein the dynamic responses of said at least two models differ by a factor of three.

11. The controller of claim 1, wherein there are provided at least three models for modeling three of the selected aspects, they being different aspects and having different dynamic responses.

12. The controller of claim 11, wherein the dynamic responses of said three models differ such that said first model is slower than second model and said second model is slower than said third model in dynamic responses associated therewith.

13. A method for controlling a multi-variable input plant having a plurality of manipulatible variables (MVs) as inputs, and operable to provide an output, comprising the steps of:

providing a predictive model of the dynamics of selected aspects of the operation of the plant for modeling the dynamics thereof, the model having at least two discrete models associated therewith that model at least two of the selected aspects, the at least two discrete models having different dynamic responses and each of the discrete models providing a predicted output of the respective selected aspect;

receiving in an optimizer desired values for the selected aspects of the operation of the plant modeled by the predictive model and the predicted outputs from the predictive model and optimizing the inputs to the model to minimize error between the predicted and desired values; and

applying the optimized input values to the plant after optimization thereof.

14. The method of claim 13, wherein the at least two discrete models are linear dynamic models.

15. The method of claim 13, wherein the at least two discrete models are independent of each other.

16. The method of claim 13, wherein the at least two discrete models are separate and independent models.

17. The method of claim 13, wherein the step of optimizing is further operable to receive constraints on the input values such that the optimization operation changes the inputs within the constraints when minimizing the error between the predicted and desired values.

18. The method of claim 13, wherein the plant is subject to at least one disturbance variable that alters the operation of the plant and wherein the predictive model will predict the behavior of the selected aspects modeled thereby in response to changes in the at least one disturbance variable and the step of optimizing will change the inputs to again minimize the error between the predicted and desired values.

19. The method of claim 13, wherein the two discrete models are linear dynamic models and wherein each the step of providing the at least two discrete models comprises:

providing a plurality of independent sub models, each associated with one of the inputs to the associated model, each of the independent sub models operable to model the associated select aspect as a function of the associated input thereto; and

summing the output of the independent sub models for each of the discrete models to provide a single output associated with each of the discrete models that is a function of all of the inputs that are input to each of the associated sub models for each of the discrete models.

20. The method of claim 19, wherein the step of optimizing comprises the steps of:

comparing with a comparator for each of the discrete models the associated desired value with the predicted output of the discrete model to generate an error value as the difference therebetween for each of the discrete models; and

incrementally changing with an iterative optimization device the MV inputs over a defined dynamic path for all of the discrete models and inputting them to the model over a plurality of iterative cycles to minimize the error between the desired values and the predicted values.

21. The method of claim 20, wherein the step of incrementally changing is operable to utilize a steepest descent optimization algorithm for the optimization procedure.

22. The method of claim 13, wherein the dynamic responses of the at least two models differ by a factor of three.

23. The method of claim 13, wherein there are provided at least three models for modeling three of the selected aspects, they being different aspects and having different dynamic responses.

24. The method of claim 23, wherein the dynamic responses of the three models differ such that the first model is slower than second model and the second model is slower than the third model in dynamic responses associated therewith.