Scalable, hierarchical control for complex processes

• US 6,810,291 B2
• Filed: 09/13/2002
• Issued: 10/26/2004
• Est. Priority Date: 09/14/2001
• Status: Expired due to Fees

• Current Assignee: IBEX Process Technology
• Original Assignee: IBEX Process Technology
• Inventors: Rietman, Edward A., Card, Jill P.
• Primary Examiner(s): Voeltz, Emanuel Todd
• Application Number:
  US10/243,963
• Publication Number:
  US 20030083757A1
• Time in Patent Office: 774 Days
• Field of Search: 700/30, 700/31, 700/33, 700/36, 700/44, 700/48, 700/49, 700/50, 700/121, 438/5, 438/7, 438/8
• US Class Current: 700/48
• CPC Class Codes: G05B 13/027 using neural networks only

Abstract

The present invention provides a method and system for complex process optimization utilizing metrics, operational variables, or both, of one or more process steps and optimization of one or more of these process step parameters with respect to a cost function for the parameter. In one embodiment, the invention provides a scalable, hierarchical optimization method utilizing optimizations at one process level as inputs to an optimization of a higher or lower process level.

Claims

1. A method of optimization of a process having an associated metric and comprising a plurality of sub-processes, the method comprising the steps of:

(a) providing a target metric value for the process;

(b) providing one or more ranges of acceptable values for the sub-process metrics to define a constraint set;

(c) providing a nonlinear regression model that has been trained in the relationship between the sub-process metrics and the process metric such that the nonlinear regression model can determine a predicted process metric value from the measured sub-process metric values; and

(d) using the nonlinear regression model and an optimizer to determine values for the sub-process metrics within the constraint set that produce at a substantially lowest cost a predicted process metric value substantially as close as possible to the target process metric value.

2. The method of claim 1, further comprising the step of:

repeating steps (a)-(d) for a sub-process of the process, wherein said sub-process becomes the process and one or more sub-sub-processes of said sub-process become the sup-processes of steps (a)-(d).

3. The method of claim 1, further comprising the step of:

repeating steps (a)-(d) for a higher level process of comprising a plurality of the processes of claim 1, wherein said higher level process becomes the process and one or more of the plurality of the processes of claim 1 become the sup-processes of steps (a)-(d).

4. The method of claim 1, wherein the optimizer associates costs with at least one of the sub-process metrics.

5. The method of claim 1, further comprising the steps of:

(e) providing one or more target sub-process metric values for a sub-process, the target sub-process metric values producing at a substantially lowest cost a predicted process metric value substantially as close as possible to the target process metric value;

(f) providing one or more ranges of acceptable values for the sub-process operational variables to define an operational variable constraint set;

(g) providing a nonlinear regression model that has been trained in the relationship between the sub-process operational variables and the sub-process metrics of said sub-process such that the nonlinear regression model can determine a predicted sub-process metric value for said sub-process from the sub-process operational variable values; and

(h) using the nonlinear regression model and an optimizer to determine target values for the sub-process operational variables within the operational variable constraint set that produce at a substantially lowest cost a predicted sub-process metric for said sub-process as close as possible to the target sub-process metric for said sub-process.

6. The method of claim 5, further comprising the step of:

repeating steps (e)-(h) for another sub-process of the process.

7. The method of claim 5, wherein the optimizer associates costs with at least one of the sub-process operational variables.

8. The method of claim 5, further comprising the steps of:

(i) measuring at least one of one or more sub-process metrics and one or more sub-process operational variables; and

(j) adjusting one or more sub-process operational variables as close as possible to a target value for the sub-process operational variable.

9. The method of claim 8, further comprising the step of:

repeating steps (i)-(j) for another sub-process of the process.

10. An article of manufacture having a computer-readable medium with computer-readable instructions embodied thereon for performing the method of claim 5.

11. The method of claim 1, wherein the action of providing of at least one of steps (a) and (b) comprises measuring.

12. An article of manufacture having a computer-readable medium with computer-readable instructions embodied thereon for performing the method of claim 1.

13. A method of optimization of a process having an associated metric and comprising a plurality of sub-processes, the method comprising the steps of:

(a) providing a target metric value for the process;

(b) providing one or more ranges of acceptable values for the sub-process metrics and the sub-process operational variables to define a sub-process constraint set;

(c) providing a nonlinear regression model that has been trained in the relationship between the sub-process metrics and sub-process operational variables and the process metric such that the nonlinear regression model can determine a predicted process metric value from the measured sub-process metric and operational variable values; and

(d) using the nonlinear regression model and an optimizer to determine values for the sub-process metrics and the sub-process operational variables within the sub-process constraint set that produce at a substantially lowest cost a predicted process metric substantially as close as possible to the target process metric.

14. The method of claim 13, wherein the optimizer associates costs with at least one of the sub-process metrics and at least one of the sub-process operational variables.

15. The method of claim 13, further comprising the step of:

repeating steps (a)-(d) for another sub-process of the process.

16. The method of claim 13, further comprising the steps of:

(e) measuring at least one of one or more sub-process metrics and one or more sub-process operational variables; and

(f) adjusting one or more sub-process operational variables as close as possible to a target value for the sub-process operational variable.

17. The method of claim 15, further comprising the step of:

repeating steps (e)-(f) for another sub-process of the process.

18. An article of manufacture having a computer-readable medium with computer-readable instructions embodied thereon for performing the method of claim 13.

19. A method of compensating for sub-process deviation from an acceptable range about a target metric, the method comprising the steps of:

(a) providing a target metric value for a metric associated with a process, the process comprising a sequence of sub-processes;

(b) providing a target sub-process metric value for a metric associated with a sub-process of the process;

(c) providing one or more ranges of acceptable values for the sub-process metrics to define a constraint set;

(d) detecting a substantial deviation in at least one sub-process metric value of one sub-process from the target sub-process metric value for said sub-process that defines a deviating sub-process;

(e) providing a nonlinear regression model that has been trained in the relationship between the sub-process metrics and the process metric such that the nonlinear regression model can determine a predicted process metric value based on the sub-process metric value of the deviating sub-process; and

(f) using the nonlinear regression model and an optimizer to determine values for the sub-process metrics of sub-processes that are later in the process sequence than the deviating sub-process, wherein said values are within the constraint set and produce at a substantially lowest cost a predicted process metric value substantially as close as possible to the target process metric value.

20. The method of claim 19, wherein step (b) comprises:

providing a target metric value for a metric associated with a process, the process comprising a sequence of sub-processes;

providing a nonlinear regression model that has been trained in the relationship between the sub-process metrics and the process metric such that the nonlinear regression model can determine a predicted process metric value based on the sub-process metric value of the deviating sub-process; and

using the nonlinear regression model and an optimizer to determine values for the sub-process metrics of sub-processes to define target sub-process metric values, wherein said values are within a constraint set and produce at a substantially lowest cost a predicted process metric value substantially as close as possible to the target process metric value.

21. The method of claim 19, wherein the values for the sub-process metrics of sub-processes that are later in the process sequence than the deviating sub-process of step (f) define compensating target sub-process metric values;

and the method further comprises the steps of;

(g) providing one or more ranges of acceptable values for one or more sub-process operational variables to define an operational variable constraint set;

(h) providing a nonlinear regression model that has been trained in the relationship between one ore more of the sub-process operational variables and the sub-process metrics of said sub-process such that the nonlinear regression model can determine a predicted sub-process metric value for said sub-process from the sub-process operational variable values; and

(i) using the nonlinear regression model and an optimizer to determine target values for one or more of the sub-process operational variables within the operational variable constraint set that produce at a substantially lowest cost a predicted sub-process metric for said sub-process as close as possible to the compensating target sub-process metric value for said sub-process.

22. The method of claim 21, further comprising the steps of:

(j) measuring at least one of one or more sub-process metrics and one or more sub-process operational variables; and

(k) adjusting one or more sub-process operational variables as close as possible to the target value for the sub-process operational variable.

23. The method of claim 19, wherein the optimizer associates costs with at least one of the sub-process operational variables.

24. An article of manufacture having a computer-readable medium with computer-readable instructions embodied thereon for performing the method of claim 19.

25. A data processing device for optimizing a process having an associated metric and comprising a plurality of sub-processes, the device receiving a target metric value for the process and one or more ranges of acceptable values for the sub-process metrics to define a constraint set, and executing a nonlinear regression model that has been trained in the relationship between the sub-process metrics and the process metric such that the nonlinear regression model can determine a predicted process metric value from the measured sub-process metric values, the device being configured to use the nonlinear regression model and an optimizer to determine values for the sub-process metrics within the constraint set that produce at a substantially lowest cost a predicted process metric value substantially as close as possible to the target process metric value.

26. A data processing device for optimizing a process having an associated metric and comprising a plurality of sub-processes, the device receiving a target metric value for the process and one or more ranges of acceptable values for the sub-process metrics and the sub-process operational variables to define a sub-process constraint set, and executing a nonlinear regression model that has been trained in the relationship between the sub-process metrics and sub-process operational variables and the process metric such that the nonlinear regression model can determine a predicted process metric value from the measured sub-process metric and operational variable values, the device being configured to use the nonlinear regression model and an optimizer to determine values for the sub-process metrics and the sub-process operational variables within the sub-process constraint set that produce at a substantially lowest cost a predicted process metric substantially as close as possible to the target process metric.

27. A data processing device for compensating for sub-process deviation from an acceptable range about a target metric, the device receiving (i) a target metric value for a metric associated with a process, the process comprising a sequence of sub-processes, (ii) a target sub-process metric value for a metric associated with a sub-process of the process, and (iii) one or more ranges of acceptable values for the sub-process metrics to define a constraint set, the device detecting a substantial deviation in at least one sub-process metric value of one sub-process from the target sub-process metric value for said sub-process that defines a deviating sub-process and comprising a nonlinear regression model that has been trained in the relationship between the sub-process metrics and the process metric such that the nonlinear regression model can determine a predicted process metric value based on the sub-process metric value of the deviating sub-process, the device being configured to use the nonlinear regression model and an optimizer to determine values for the sub-process metrics of sub-processes that are later in the process sequence than the deviating sub-process, wherein said values are within the constraint set and produce at a substantially lowest cost a predicted process metric value substantially as close as possible to the target process metric value.