Analyzer for modeling and optimizing maintenance operations

• US 6,110,214 A
• Filed: 08/23/1996
• Issued: 08/29/2000
• Est. Priority Date: 05/03/1996
• Status: Expired due to Term

• Current Assignee: Aspentech Corporation
• Original Assignee: Aspen Technology (Telefonaktiebolaget LM Ericsson)
• Inventors: Klimasauskas, Casimir C.
• Primary Examiner(s): Teska, Kevin J.
• Assistant Examiner(s): FREJD, RUSSELL WARREN
• Application Number:
  US08/702,148
• Time in Patent Office: 1,467 Days
• Field of Search: 395/500, 395/650, 395/500.23, 364/148, 364/149, 364/551.01, 364/138, 700/29, 700/28, 700/31, 700/48, 700/30, 706/16, 706/23
• US Class Current: 703/2
• CPC Class Codes: G05B 13/027 using neural networks only
• CPC Class Codes: G05B 13/0275 using fuzzy logic only
• CPC Class Codes: G05B 19/042 using digital processors G05B19/05 takes precedence

Abstract

A first model or first analyzer having a series of filters is provided to represent time-varying effects of maintenance events. The first model or analyzer further enhances the selection of derived variables which are used as inputs to the first analyzer. Additionally, a combination of fuzzy logic and statistical regression analyzers are provided to better model the equipment and the maintenance process. An optimizer with a bi-modal optimization process which integrates discrete maintenance events with continuous process variables is also provided. The optimizer determines the time and the type of maintenance activities which are to be executed, as well as the extent to which the maintenance activities can be postponed by changing other process variables. Thus, potential modifications to process variables are determined to improve the current performance of the processing equipment as it drifts out of tolerance.

Claims

1. A method for modeling a maintenance task having input parameters and one or more activities, said method comprising the steps of:

collecting said input parameters;

generating derived variables, including the steps of;

performing a search on said input parameters using one of more functional forms;

correlating said functional forms to the results of the performed search, and

ranking said functional forms based on the results of the step of correlating; and

applying a primary analyzer to said input parameters and said derived variables to generate outputs corresponding to said activities.

2. The method of claim 1, wherein said collecting step further comprises the steps of receiving and modifying time-varying input parameters.

3. The method of claim 2, wherein said modifying step filters said time-varying input parameters.

4. The method of claim 2, wherein said collecting step collects operating periods between said activities.

5. The method of claim 1, wherein said collecting step encodes maintenance type events.

6. The method of claim 1, wherein said collecting step filters said input parameters.

7. The method of claim 1, wherein said correlating step comprises the step of applying a uniformly distributed sub-set of examples of said input parameters.

8. The method of claim 1, wherein said applying step applies a partial least squares analyzer having a linear inner model.

9. The method of claim 1, wherein said applying step applies a partial least squares analyzer having a fuzzy single-input-single-output (FSISO) model.

10. The method of claim 1, wherein said applying step applies a partial least squares analyzer having a kernel density estimator.

11. The method of claim 1, wherein said applying step applies a partial least squares analyzer having a generalized regression neural network.

12. The method of claim 1, further comprising the step of optimizing said analyzer.

13. The method of claim 12, wherein said input parameters include continuous variables and wherein said optimizing step further comprises the step of integrating discrete events representative of said activities with said continuous variables.

14. The method of claim 12, wherein said optimizing step further comprises the steps of:

generating all combinations of said discrete activities into enumerated sets; and

deriving a common objective function for each of said enumerated sets, including the steps of;

evaluating the cost of said set of discrete activities;

optimizing said continuous activities based on said cost; and

merging the results of said evaluating step and said continuous activities optimizing step into said common objective function.

15. The method of claim 1, wherein said task is a plasma etching equipment maintenance task.

16. A method for modeling a task having input parameters and one or more activities, said method comprising the steps of:

collecting time-varying input parameters;

modifying said time-varying input parameters;

generating derived variables, including the steps of;

performing a search on said time-varying input parameters using one or more functional forms;

correlating said functional forms to said search results; and

ranking said functional form based on said correlated results;

applying a primary analyzer to said input parameters to generate outputs corresponding to said activities; and

optimizing said outputs, including the steps of;

generating all combinations of said discrete activities into enumerated sets; and

deriving a common objective function for each of said enumerated sets, including the steps of;

evaluating the cost of said set of discrete activities;

optimizing said continuous activities based on said cost; and

merging the results of said evaluating step and said continuous activities optimizing step into said common objective function.

17. The method of claim 16, wherein said primary analyzer applying step applies a partial least squares analyzer having a fuzzy single-input-single-output (FSISO) model.

18. The method of claim 16, wherein said primary analyzer applying step applies a partial least squares analyzer having a kernel density estimator.

19. An analyzer for an equipment having one or more input parameters and activities, said analyzer controlling one or more task activities directed at said equipment, said analyzer comprising:

a data collection device coupled to said equipment for collecting input parameters from said equipment; and

a primary analyzer coupled to said data collection device, said primary analyzer generating outputs relating to activities to be performed on said equipment.

20. The analyzer of claim 19, wherein said data collection device further comprises one or more filters adapted to receive said input parameters and to compensate for time-varying input parameters.

21. The analyzer of claim 19, further comprising a derived variable generator coupled to said data collection device, said derived variable generator including:

a search engine coupled to said data collection device, said search engine examining said input parameters using one or more functional forms;

a correlator coupled to said search engine; and

a sorter coupled to said correlator, said sorter ranking said functional forms based on the result of said correlator.

22. The analyzer of claim 19, wherein said primary analyzer is a partial least squares (PLS) analyzer having a linear inner model.

23. The analyzer of claim 19, wherein said primary analyzer is a partial least squares analyzer having a fuzzy SISO model.

24. The analyzer of claim 19, wherein said primary analyzer is a partial least squares analyzer having a polynomial analyzer.

25. The analyzer of claim 24, wherein said primary analyzer is a group method of data handling (GMDH) polynomial analyzer.

26. The analyzer of claim 19, wherein said primary analyzer is a partial least squares analyzer having a neural network.

27. The analyzer of claim 19, wherein said primary analyzer is a partial least squares analyzer having a hybrid linear-neural network.

28. The analyzer of claim 19, wherein said primary analyzer is a partial least squares analyzer having a kernel density estimator.

29. The analyzer of claim 19, wherein said primary analyzer is a partial least squares analyzer having a generalized regression neural network.

30. The analyzer of claim 19, further comprising an optimizer coupled to said primary analyzer, said optimizer including:

an enumerator for generating all combinations of discrete activities into enumerated sets; and

a common objective function generator coupled to said enumerator, said common objective function generator fixer including;

a cost analyzer for evaluating the cost of said discrete activities;

a continuous activity optimizer coupled to said cost analyzer; and

an objective generator coupled to said cost analyzer and said continuous activity optimizer, said objective generator merging the outputs of said cost analyzer and said continuous activity optimizer into said common objective function.

31. The analyzer of claim 19, wherein said equipment is a plasma etching equipment.

32. An analyzer for a plasma etching equipment having one or more input parameters and activities, said input parameters having discrete and continuous variables, said analyzer controlling one or more task activities directed at said equipment, said analyzer comprising:

a data collection device coupled to said equipment for collecting input parameters from said equipment;

a derived variable generator, including;

a search engine coupled to said data collection device, said search engine examining said input parameters using one or more functional forms;

a correlator coupled to said search engine; and

a sorter coupled to said correlator, said sorter ranking said functional forms based on the result of said correlator;

a primary analyzer coupled to said data collection device, said primary analyzer generating outputs relating to activities to be performed on said equipment; and

an optimizer, including;

an enumerator for generating all combinations of discrete activities into enumerated sets; and

a common objective function generator coupled to said enumerator, said common objective function generator further including;

a cost analyzer for evaluating the cost of said discrete activities;

a continuous activity optimizer coupled to said cost analyzer; and

an objective generator coupled to said cost analyzer and said continuous activity optimizer, said objective generator merging the outputs of said cost analyzer and said continuous activity optimizer into said common objective function.

33. The analyzer of claim 32, wherein said primary analyzer is a partial least squares analyzer having a fuzzy SISO model.

34. A method for modeling a task having input parameters and one or more activities, said method comprising the steps of:

collecting said input parameters;

generating derived variables, including the steps of;

performing a search on said input parameters using one of more functional forms;

correlating said functional forms to the results of the performed search;

ranking said functional forms based on the results of the step of correlating;

applying a primary analyzer to said input parameters and said derived variables to generate outputs corresponding to said activities; and

optimizing said analyzer.

35. The method of claim 34, wherein said input parameters include continuous variables and wherein said optimizing step further comprises the step of integrating discrete events representative of said activities with said continuous variables.

36. The method of claim 34, wherein said optimizing step further comprises the steps of:

generating combinations of said discrete event activities into enumerated sets; and

deriving a common objective function for each of said enumerated sets, including the steps of;

evaluating the cost of said set of discrete event activities;

optimizing said continuous activities based on said cost; and

merging the results of said evaluating step and said continuous activities optimizing step into said common objective function.

37. A method for modeling a maintenance task having input parameters and one or more maintenance activities, said method comprising the steps of:

collecting said input parameters for the task;

applying a primary analyzer to said input parameters to generate outputs corresponding to said activities.

38. The method of claim 37, wherein the primary analyzer has a series of filters to represent time-varying effects of maintenance events.

39. A method for modeling a task having input parameters and one or more activities, said method comprising the steps of:

collecting said input parameters;

applying a primary analyzer to said input parameters to generate outputs corresponding to said activities; and

optimizing said analyzer.

40. The method of claim 39, wherein said input parameters include continuous variables and wherein said optimizing step further comprises the step of integrating discrete events representative of said activities with said continuous variables.

41. The method of claim 40, wherein said optimizing step further comprises the steps of:

generating all combinations of said discrete activities into enumerated sets; and

deriving a common objective function for each of said enumerated sets, including the steps of;

evaluating the cost of said set of discrete activities;

optimizing said continuous activities based on said cost; and

merging the results of said evaluating step and said continuous activities optimizing step into said common objective function.