Adaptive object-oriented optimization software system

• US 6,112,126 A
• Filed: 02/20/1998
• Issued: 08/29/2000
• Est. Priority Date: 02/21/1997
• Status: Expired due to Fees

• Current Assignee: Lmhales, LLC
• Original Assignee: Baker Hughes Incorporated (Baker Hughes Company)
• Inventors: Hales, Lynn B., Ynchausti, Randy A., Foot, Donald G. Jr.
• Primary Examiner(s): Gordon, Paul P.
• Assistant Examiner(s): PATEL, RAMESH B
• Application Number:
  US09/027,339
• Time in Patent Office: 921 Days
• Field of Search: 700/29, 700/28, 700/30, 700/31, 700/32, 700/49, 395/500.23, 395/702, 395/704, 395/710, 703/2, 717/2, 717/4, 717/10
• US Class Current: 700/29
• CPC Class Codes: E21B 2200/22 Fuzzy logic, artificial intelligence, neural networks or the like
• CPC Class Codes: G05B 13/029 using neural networks and expert systems

Abstract

The present invention relates to process control optimization systems which utilize an adaptive optimization software system comprising goal seeking intelligent software objects; the goal seeking intelligent software objects further comprise internal software objects which include expert system objects, adaptive models objects, optimizer objects, predictor objects, sensor objects, and communication translation objects. The goal seeking intelligent software objects can be arranged in a hierarchical relationship whereby the goal seeking behavior of each intelligent software object can be modified by goal seeking intelligent software objects higher in the hierarchical structure. The goal seeking intelligent software objects can also be arranged in a relationship which representationally corresponds to the controlled process'"'"' flow of materials or data.

Claims

1. A goal seeking intelligent software object, useful in process control systems, capable of influencing processes controlled by said process control system through a plurality of encountered states wherein

said goal seeking intelligent software object adapts to said controlled process by creating, selecting, learning, training, and remembering one or more predictive software models for each encountered state of said goal seeking intelligent software object whereby

for each of said encountered states of said process for which more than one said predictive software model exists, said goal seeking intelligent software object selects one or more of said predictive software models, previously trained, to model said controlled process.

2. A goal seeking intelligent software object comprising a plurality of internal software objects wherein

said internal software objects are each in communication with one another,

said internal software objects exhibit individualized goal seeking behavior,

said internal software objects interact with one another to exhibit aggregate goal seeking behavior, and

one or more of said internal software objects may modify said individualized goal seeking behavior of one or more of said other internal software objects.

3. A goal seeking intelligent software object comprising:

a plurality of internal software objects, said plurality of internal software objects further comprising

at least one expert system software object,

at least one adaptive models software object,

at least one predictor software object, and

at least one optimizer software object wherein

said expert system software object is in communication with said optimizer software object and can modify said optimizer software object'"'"'s behavior,

said expert system software object is in communication with said predictor software object and can modify said predictor software object'"'"'s behavior,

said expert system software object is in communication with said adaptive models software object and can modify said adaptive models software object'"'"'s behavior,

said optimizer software object is in communication with said expert system software object and can modify said expert system software object'"'"'s behavior,

said optimizer software object is in communication with said predictor software object and can modify said predictor software object'"'"'s behavior, and

said optimizer software object is in communication with said adaptive models software object and can modify said adaptive models software object'"'"'s behavior whereby

said plurality of internal software objects exhibit aggregate goal seeking behavior.

4. The goal seeking intelligent software object as described in claim 3 wherein said expert system software object further comprises a rules knowledge base comprising one or more rules selected from a set of crisp logic rules and fuzzy logic rules.

5. The goal seeking intelligent software object as described in claim 3 wherein said optimizer software object further comprises optimization methodologies wherein said optimization methodologies include one or more genetic algorithms.

6. The goal seeking intelligent software object as described in claim 3 further comprising modeling methodologies wherein said modeling methodologies include one or more neural networks.

7. The goal seeking intelligent software object as described in claim 3 wherein said communication translator software object further comprises communications protocols selected from a set comprising internal protocols, protocols required by devices external to said goal seeking intelligent software object, serial data communication protocols, parallel data communication protocols, local area network data communication protocols, and wide area data communication protocols.

8. A goal seeking intelligent software object which comprises:

at least one expert system software object comprising at least one rules knowledge base;

at least one adaptive models software object generating one or more predictive models from one or more input values and from one or more modeling methodologies;

at least one predictor software object selecting an optimal one of said predictive models using predictor selection criteria and optimizing methodologies;

at least one optimizer software object setting optimal output data values for said optimizer software object utilizing one or more objective goals and one or more optimization methodologies;

at least one communication translator software object communicating data using one or more data communications protocols; and

at least one sensor software object which may accept data from said expert system software objects, said optimizer software objects, said adaptive models software objects, said predictor software objects, said communication translator software objects, and other sensor software objects;

process said data for storage;

store said data; and

provide stored data on request by said expert system software objects, said optimizer software objects, said adaptive models software objects, said predictor software objects, said communication translator software objects, and other sensor software objects to said expert system software objects, said optimizer software objects, said adaptive models software objects, said predictor software objects, and said communication translator software objects;

whereby

said expert system software object is in communication with said optimizer software object and can modify said optimizer software object said objective goals,

said expert system software object is in communication with said optimizer software object and can modify said optimization methodologies,

said expert system software object is in communication with said predictor software object and can modify said predictor selection criteria of said predictor software object,

said expert system software object is in communication with said adaptive models software object and can modify said modeling methodologies of said adaptive models software object,

said optimizer software object is in communication with said expert system software object and can modify said rules knowledge base of said expert system software object,

said optimizer software object is in communication with said predictor software object and can modify said selection criteria of said predictor software object, and

said optimizer software object is in communication with said adaptive models software object and can modify said modeling methodologies of said adaptive models software object.

9. A goal seeking intelligent software object as in claim 8 further comprising a plurality of said sensor software objects, each said sensor software object additionally able to provide said stored data on request by another of said sensor software objects.

10. An adaptive optimization software system comprising:

a plurality of goal seeking intelligent software objects, useful in process control systems, capable of influencing processes controlled by said process control system through a plurality of encountered states wherein

said goal seeking intelligent software objects adapt to said controlled process by creating, selecting, learning, training, and remembering one or more predictive software models for each encountered state of said goal seeking intelligent software objects whereby

for each of said encountered states of said process for which more than one said predictive software model exists, said goal seeking intelligent software objects select predictive software models, previously trained, to model said controlled process,

said plurality of said goal seeking intelligent software objects relationally communicate with other of said goal seeking intelligent software objects, and

said goal seeking intelligent software objects may each communicate with devices external to any of said goal seeking intelligent software objects.

11. An adaptive optimization software system as described in claim 10 further comprising a graphical user interface wherein said graphical user interface is used to define said relational communication from a set of relational communications consisting of a user defined number of hierarchical relationships and flow relationships, whereby

said hierarchical relationships define prioritization and scope relationships between a plurality of said goal seeking intelligent software objects, and

said flow relationships representationally correspond to data communicated between said plurality of said goal seeking intelligent software objects.

12. An adaptive optimization software system as described in claim 11 wherein said prioritization and scope of said goal seeking intelligent software object at the highest hierarchical relationship level are economics based.

13. An adaptive optimization software system as described in claim 11 wherein said prioritization and scope of said goal seeking intelligent software object at the lowest hierarchical relationship level are non-economics based.

14. A process control optimization system comprising:

a computer system further comprising

one or more display devices comprising one or more display areas, and end user input devices;

a controllable process further comprising

external devices further comprising

one or more controllable devices wherein

said controllable devices provide interfaces which provide communication means between said computer system and said controllable devices; and

instrumentation providing data; and

one or more optimization goals for said process control system; and

an adaptive optimization software system further comprising a plurality of goal seeking intelligent software objects capable of influencing said controllable process wherein

said goal seeking intelligent software objects further comprise

one or more rules knowledge databases,

one or more modeling methodologies,

one or more predictor selection criteria, and

one or more optimal objective goals;

said goal seeking intelligent software objects adapt to said controllable process by creating, selecting, learning, training, and remembering one or more predictive software models for each encountered state of said goal seeking intelligent software objects; and

one or more said goal seeking intelligent software objects are in communication with one or more of said external devices whereby

for each of said encountered states of said controllable process for which more than one said predictive model exists, said goal seeking intelligent software objects select predictive software models, previously trained, to model said controllable process wherein

one or more of said rules knowledge databases comprises rules to implement said optimization goals for said controllable process,

one or more of said modeling methodologies generate models of said controllable process,

one or more of said predictor selection criteria exemplify said optimization goals for said controllable process, and

one or more of said optimization goals for said controllable process include optimization of said process control system;

said plurality of said goal seeking intelligent software objects relationally communicate with other of said goal seeking intelligent software objects; and

each of said goal seeking intelligent software objects in communication with one of said external devices representationally corresponds to that said external device.

15. A process control optimization system as described in claim 14 wherein said instrumentation providing data provides actual real-world device data.

16. A process control optimization system as described in claim 14 wherein said instrumentation providing data provides simulated real-world device data.

17. A process control optimization system as described in claim 14 further comprising a graphical user interface wherein said graphical user interface is used to define any number of said relational communications from a set of relational communications consisting of hierarchical relationships and flow relationships, whereby

said hierarchical relationships define prioritization and scope relationships between said goal seeking intelligent software objects,

said flow relationships representationally correspond to the process which is to be controlled by said goal seeking intelligent software objects, and

said flow relationships further representationally correspond to data communicated between said goal seeking intelligent software objects.

18. A method of adaptive optimization of a process controlled optimization system wherein said process control optimization system provides a plurality of goal seeking intelligent software objects further comprising sensor software objects providing current data, historical data, and statistical data;

expert system software objects providing one or more associated rules knowledge bases;

adaptive models software objects providing one or more modeling methodologies;

predictor software objects providing one or more predictor selection criteria;

optimizer software objects providing one or more goals and objective functions, and process constraints; and

communications translator software objects providing one or more data communications protocols for a given sampling delta, comprising the concurrent steps of;

conducting a process which is controlled by said process controlled optimization system;

determining, within said optimizer software objects, optimal output data values which best achieve said goals and objective functions without violating said process constraints;

examining, within said expert system software objects, said best fit predictive models that achieve said goals and objective functions without violating said process constraints; and

determining, within said expert system software objects, appropriate adaptive interventions.

19. A method of adaptive optimization of a process controlled optimization system as in claim 18 wherein said process control optimization system is used to simulate a real-world process useful for training end users.

20. A method of adaptive optimization of a process controlled optimization system as in claim 18 wherein said determining optimal output data values step further comprises the steps of

providing said optimizer software objects with one or more input values and with said current data, historical data, and statistical data,

determining, within said optimizer software objects, trial values which best achieve said goals and objective..functions without violating said process constraints,

evaluating, within said optimizer software objects, said trial values, said evaluating step further comprising the steps of

determining, within said predictor software objects, best fit predictive models, said best fit predictive models most closely matching said output data values;

simulating, calculating, and predicting, within said adaptive models software objects associated with said optimizer software objects, future performance of said process which is controlled using said trial values and said best file predictive model;

determining, within said optimizer software objects, whether said best fit predictive models achieve said goals and objective functions without violating said process constraints;

retaining, within said optimizer software objects, said trial values if said best fit predictive models achieve said goals and objective functions without violating said process constraints; and

attempting, within said optimizer software objects, a new set of trial values if said best fit predictive models do not achieve said goals and objective functions without violating said process constraints.

21. A method of adaptive optimization of a process controlled optimization system as in claim 18 wherein said determining appropriate adaptive interventions step further comprises the steps of

utilizing, within said expert system software objects, one or more of said rules knowledge bases to adaptively modify performance goals for said optimizer software objects;

utilizing, within said expert system software objects, one or more of said rules knowledge bases to adaptively modify said optimizer software objects'"'"' configuration;

utilizing, within said expert system software objects, one or more of said rules knowledge bases to adaptively modify performance goals for said adaptive models software objects;

utilizing, within said expert system software objects, one or more said rules knowledge bases to adaptively modify said adaptive models software objects'"'"' configuration;

utilizing, within said expert system software objects, one or more of said rules knowledge bases to adaptively modify performance goals for said predictor software objects; and

utilizing, within said expert system software objects, one or more of said rules knowledge bases to adaptively modify said predictor software objects'"'"' configuration.