Method and system for performing model-based multi-objective asset optimization and decision-making

• US 7,536,364 B2
• Filed: 04/28/2005
• Issued: 05/19/2009
• Est. Priority Date: 04/28/2005
• Status: Active Grant

• Current Assignee: Baker Hughes, a GE Company, LLC (Baker Hughes Company)
• Original Assignee: General Electric Company
• Inventors: Schmid, James J., Bonissone, Piero P., Iyer, Naresh S., Eklund, Neil H., Knodle, Chad E., Subbu, Rajesh V., Shah, Rasiklal P., Yan, Weizhong
• Primary Examiner(s): Vincent; David R
• Assistant Examiner(s): Brown, Jr.; Nathan H
• Application Number:
  US11/116,920
• Publication Number:
  US 20060271210A1
• Time in Patent Office: 1,482 Days
• Field of Search: 706/13, 706/15, 700 28- 34
• US Class Current: 706/13
• CPC Class Codes: G06F 18/2113 by ranking or filtering the set of features, e.g. using a measure of variance or of feature cross-correlation
• CPC Class Codes: G06N 3/126 Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Abstract

A method and system for performing model-based multi-objective asset optimization and decision-making is provided. The method includes building at least two predictive models for an asset. The building includes categorizing operational historical data via at least one of: controllable variables, uncontrollable variables, output objectives, and constraints. The building also includes selecting at least two output objectives or constraints, and identifying at least one controllable or uncontrollable variable suitable for achieving the at least two output objectives or constraints. The method also includes validating each predictive model and performing multi-objective optimization using the predictive models. The multi-objective optimization includes specifying search constraints and applying a multi-objective optimization algorithm. The method further includes generating a Pareto Frontier, and selecting a Pareto optimal input-output vector.

Claims

1. A computerized method for performing multi-objective asset optimization and decision-making using predictive modeling, comprising:

building, via a process manager application executing on a processor, at least two predictive models for an asset, the asset comprising a physical machine that is communicatively coupled to the processor, the building comprising;

categorizing operational historical data of the asset that is retrieved from a storage device, the operation historical data categorized by at least one of;

controllable variables;

uncontrollable variables;

output objectives; and

constraints;

selecting at least two output objectives or constraints; and

identifying at least one controllable or uncontrollable variable suitable for achieving the at least two output objectives or constraints;

inputting, via the process manager application, the at least one controllable or uncontrollable variable to each of the at least two predictive models;

validating, via the process manager application, each predictive model;

if results of the validating indicate a confidence level above a specified threshold, applying, via the process manager application, a live data stream of inputs from the asset to the predictive models;

if results of the validating indicate a confidence level at or below a specified threshold, selecting, via the process manager application, at least one alternative controllable or uncontrollable variable for input to the predictive models;

performing, via the process manager application, multi-objective optimization using the predictive models, comprising;

specifying search constraints, comprising;

upper and lower bounds for each input variable; and

tolerance levels representing a range of values for achieving optimal output objectives, and constraints;

applying a multi-objective optimization algorithm; and

generating a Pareto Frontier, the Pareto Frontier including optimal input-output vectors;

using results of the multi-objective optimization, selecting, via the process manager application, from the Pareto Frontier, a Pareto optimal input-output vector for deployment to the asset, the selected Pareto optimal input-output vector specifying an optimal operational state for the asset; and

re-configuring the asset, via the process manager application, using the Pareto optimal input-output vector to realize the optimal operational state.

2. The computerized method of claim 1, wherein the controllable variable includes input parameters that are changeable and the uncontrollable variable includes input parameters that are static.

3. The computerized method of claim 1, wherein the physical machine includes at least one of:

a gas turbine;

an oil-fired boiler;

a refinery boiler;

an aircraft engine;

a marine engine;

a gasoline engine;

a diesel engine; and

a hybrid engine.

4. The computerized method of claim 1, wherein the output objective includes a target goal for a process.

5. The computerized method of claim 1, wherein the constraint includes a restriction placed on an output.

6. The computerized method of claim 1, wherein the validating is accomplished by comparing predicted results of the inputting to actual results of inputting.

7. The computerized method of claim 1, wherein the multi-objective optimization algorithm is at least one of:

an evolutionary algorithm;

a gradient-based algorithm;

a hybrid algorithm; and

Pareto Frontier generator.

8. The computerized method of claim 1, further comprising selecting, via the process manager application, one of the Pareto optimal input-output vectors for deployment that minimally perturbs the asset from its existing controllable variables configuration.

9. The computerized method of claim 1, further comprising:

applying, via the process manager application, a decision function to results of the generating a Pareto Frontier, wherein selecting an input-output vector for deployment is based upon results of the applying a decision function capable of reducing the number of input-output vectors.

10. A system for performing multi-objective asset optimization and decision-making using predictive modeling, comprising:

a processor;

an asset in communication with the processor, the asset comprising a physical machine; and

a process manager application implemented by the processor, the process manager application performing;

building at least two predictive models for the asset, comprising;

categorizing operational historical data of the asset that is retrieved from a storage device, the operation historical data categorized by at least one of;

controllable variables;

uncontrollable variables;

output objectives; and

constraints;

selecting at least two output objectives or constraints; and

identifying at least one controllable or uncontrollable variable suitable for achieving the at least two output objectives or constraints;

inputting the at least one controllable or uncontrollable variable to each of the at least two predictive models;

validating each predictive model;

if results of the validating indicate a confidence level above a specified threshold, applying a live data stream of inputs from the asset to the predictive models;

if results of the validating indicate a confidence level at or below a specified threshold, selecting at least one alternative controllable or uncontrollable variable for input to the predictive models;

performing multi-objective optimization using the predictive models, comprising;

specifying search constraints, comprising;

upper and lower bounds for each input variable; and

tolerance levels representing a range of values for achieving optimal output objectives, and constraints;

applying a multi-objective optimization algorithm; and

generating a Pareto Frontier, the Pareto Frontier including optimal input-output vectors;

using results of the multi-objective asset optimization, selecting, from the Pareto Frontier, a Pareto optimal input-output vector for deployment to the asset, the Pareto optimal input-output vector specifying an optimal operational state for the asset; and

re-configuring the asset using the Pareto optimal input-output vector to realize the optimal operational state.

11. The system of claim 10, wherein the controllable variable includes input parameters that are changeable and the uncontrollable variable includes input parameters that are static.

12. The system of claim 10, wherein the physical machine includes at least one of:

a gas turbine;

an oil-fired boiler;

a refinery boiler;

an aircraft engine;

a marine engine;

a gasoline engine;

a diesel engine; and

a hybrid engine.

13. The system of claim 10, wherein the output objective includes a target goal for a process.

14. The system of claim 10, wherein the constraint includes a restriction placed on an output.

15. The system of claim 10, wherein the validating is accomplished by comparing predicted results of the inputting to actual results of inputting.

16. The system of claim 10, wherein the multi-objective optimization algorithm is at least one of:

an evolutionary algorithm;

a gradient-based algorithm;

a hybrid algorithm; and

Pareto Frontier generator.

17. The system of claim 10, wherein the process manager application further performs selecting one of the Pareto optimal input-output vectors for deployment that minimally perturbs the asset from its existing controllable variables configuration.

18. The system of claim 10, wherein the process manager application further performs:

applying a decision function to results of the generating a Pareto Frontier, wherein selecting an input-output vector for deployment is based upon results of the applying a decision function capable of reducing the number of input-output vectors.