Method for developing a unified quality assessment and providing an automated fault diagnostic tool for turbine machine systems and the like

• US 6,973,396 B1
• Filed: 05/28/2004
• Issued: 12/06/2005
• Est. Priority Date: 05/28/2004
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Iasillo, Robert J., Shah, Minesh, Singh, Ajai, Norman, Bruce, Rao, Kotesh K.
• Primary Examiner(s): Barlow, John
• Assistant Examiner(s): Walling, Meagan S.
• Application Number:
  US10/855,985
• Publication Number:
  US 20050267702A1
• Time in Patent Office: 557 Days
• Field of Search: 702/81, 702/82, 702/84, 702/179, 702/182, 735/70, 736/49, 736/58, 736/60
• US Class Current: 702/81
• CPC Class Codes: G05B 23/0229 knowledge based, e.g. expert systems; genetic algorithms

Abstract

A computer implemented process is provided for assessing and characterizing the degree of success or failure of an operational event of a machine system such as a fluid compressor machine or turbine machine or the like on a continuous numerical scale. The computer implemented process develops and tracks machine unit signatures, machine site signatures and machine fleet signatures to evaluate various operational events and provide fault detection. At least some sensor data acquired from the machine system during an operational event is transformed to correct or at least reduce variabilities in the data caused by ambient conditions and fuel quality. The transformed data is then analyzed using statistical methods to determine how closely the operational event conforms to an expected normal behavior and the information is used to develop a single comprehensive quality assessment of the event. By saving, tracking and updating operational event assessments over time, machine/component degradation may be recognized at any early stage and corrective action may be initiated in advance of a catastrophic failure.

Claims

1. A method for developing a unified quality assessment of a machine system based on sensor data characterizing one or more operational events of the machine system, at least some of the sensor data having associated correction parameters, the method comprising:

acquiring sensor data corresponding to a particular operational event;

developing operational event specific signatures from parameter plots based upon acquired sensor data, at least some of the event signatures corresponding to sensor data from parameter plots corrected by utilizing one or more corrected parameter coefficients, wherein said corrected parameter coefficients reduce or eliminate variabilities in the sensor data caused by ambient operating conditions and/or fuel type or fuel quality;

classifying signatures from both corrected and uncorrected parameter plots into one of a plurality of quality assessment categories based upon a predetermined degree of statistical correspondence between a signature and a pre-determined value or range of values; and

combining quality assessment evaluations of signatures corresponding to both corrected and uncorrected parameter plots to develop a single comprehensive quality assessment value indicative of the machine system operation in response to said one or more operational events.

2. The method of claim 1 wherein at least some acquired sensor data is corrected by applying a mathematical transformation which converts the data into a corrected parameter space.

3. The method of claim 1 wherein said plurality of quality assessment categories include at least one category indicative of an acceptable degree of correspondence, one category indicative of an marginally acceptable degree of correspondence and one category indicative of an unacceptable degree of correspondence.

4. The method of claim 1 further comprising:

saving a comprehensive quality assessment signature indicative of the machine system operation in response to an operational event.

5. The method of claim 1 further comprising:

developing an updated unified quality assessment value upon acquiring sensor data during each similar subsequent operational event.

6. The method of claim 1 further comprising:

identifying a predetermined amount of deviation of said updated quality assessment value from a previous or historical quality assessment value as indicative of a potential system or component failure.

7. The method of claim 1 wherein the machine system is a gas turbine system.

8. The method of claim 1 wherein the machine system is a wind turbine power generation system.

9. The method of claim 1 wherein the machine system is a fluid pumping system.

10. A method implemented on a computer for developing a comprehensive unified quality assessment and fault diagnostic of a machine system operational event, the method comprising:

acquiring sensor data corresponding to a particular operational event of the machine system;

performing a mathematical transformation on at least some of the acquired sensor data using a predetermined set of parameter correction coefficients such that known variabilities in the acquired sensor data are reduced or eliminated;

comparing acquired sensor data to a pre-determined acceptable value or range of acceptable values and determining an amount of correspondence between the acquired sensor data and said pre-determined value or range of values to within a computed statistical degree;

classifying both transformed and non-transformed sensor data into one of a plurality of quality assessment categories based upon said computed statistical degree of correspondence;

combining statistical evaluations of both transformed and non-transformed data into a single unified quality assessment value; and

identifying a predetermined amount of deviation of said quality assessment value from a previously determined historical quality assessment values as indicative of a potential system or component failure.

11. The method of claim 10 further comprising:

developing a new unified quality assessment value after acquiring sensor data during a subsequent operational event, said new unified quality assessment value based at least in part on previously acquired data.

12. The method of claim 10 wherein said known variabilities in the acquired sensor data are caused by ambient operating conditions or fuel quality.

13. The method of claim 10 wherein said plurality of quality assessment categories include at least one category indicative of an acceptable degree of correspondence, one category indicative of an marginally acceptable degree of correspondence and one category indicative of an unacceptable degree of correspondence.

14. The method of claim 10 wherein the machine system is a wind turbine power generation system.

15. The method of claim 10 wherein the machine system is a gas turbine system.

16. The method of claim 10 wherein the machine system is an oil or gas pumping system.

17. Computer-readable medium having computer-executable instructions of performing a method for developing a quality assessment of a turbine system operational event, the method comprising:

acquiring sensor data corresponding to a particular operational event of a turbine system;

performing a mathematical transformation on at least some of said acquired sensor data using a predetermined set of parameter corrections such that variabilities in said acquired sensor data caused by ambient operating conditions or fuel type/quality are reduced or eliminated;

comparing said acquired sensor data to a pre-determined acceptable value or range of acceptable values and determining an amount of correspondence between the acquired sensor data and said pre-determined value or range of values to within a computed statistical degree; and

classifying both transformed and non-transformed sensor data into one of a plurality of quality assessment categories based upon said computed statistical degree of correspondence; and

combining both transformed and non-transformed data into a single unified quality assessment value.

18. The computer-readable medium of claim 17 having computer-executable instructions for performing a method for developing a quality assessment of a turbine system operational event, the method further comprising:

saving the unified quality assessment value;

computing a new unified quality assessment value upon acquiring sensor data from a subsequent operational event; and

identifying a predetermined amount of deviation of said quality assessment value from a previously computed quality assessment value or range of values as indicative of a potential system or component failure.

19. The computer-readable medium of claim 17 having computer-executable instructions for performing a method for developing a quality assessment of a turbine system operational event, wherein the method further includes computer-executable instructions for saving a plurality of unified quality assessment values and automatically identifying any deviation from previously saved values which exceeds a predetermined amount as indicative of a potential system or component failure.

20. The computer-readable medium of claim 17 having computer-executable instructions for performing a method for developing a quality assessment of a turbine system operational event, wherein said plurality of quality assessment categories include at least one category indicative of an acceptable degree of correspondence, one category indicative of an marginally acceptable degree of correspondence and one category indicative of an unacceptable degree of correspondence.

21. A method implemented on a computer for developing a quality assessment of a turbine machine event based on sensor data characterizing the operation of the turbine during the operational event, the sensor data being influenced by ambient operating conditions and known variability in fuel type/quality, the method comprising:

acquiring turbine sensor data corresponding to a particular turbine operational event;

correcting at least some of the sensor data, wherein variability in the data due to variations in ambient operation conditions and fuel type/quality are reduced or eliminated; and

comparing corrected sensor data to a predetermined value or range of values; and

classifying said data into one of a plurality of quality assessment categories according to a predetermined degree of statistical correspondence between said data and the predetermined value or range of values.

22. The method of claim 21 wherein at least some acquired sensor data is corrected by a mathematical transformation which converts said data into a corrected parameter space.

23. The method of claim 21 further comprising:

comparing non-transformed sensor data and transformed data to a pre-determined expected value to assess the degree to which the data matches the pre-determined expected value; and

developing a comprehensive operational event assessment based on said comparison.

24. The method of claim 21 further comprising:

storing the developed comprehensive event assessment and performing a updated quality assessment each time additional or new sensor data is acquired.

25. A method implemented on a computer for developing a comprehensive unified quality assessment of a gat turbine system operational event, the method comprising:

acquiring turbine sensor data which characterizes the operation of the turbine during occurrence of an operational event, wherein the acquired sensor data is affected by ambient operating conditions of the turbine and/or the fuel quality/type;

using a predetermined mathematical transform or a set of correction parameters to transform acquired sensor data to effectively remove or reduce variability in the acquired sensor data which results from variations in ambient operating conditions at the turbine and/or fuel type/quality;

comparing both transformed and non-transformed sensor data relevant to the operational event with a predetermined expected value or range of values and determining a statistical degree to which said transformed data and non-transformed sensor data matches the expected value or range of values;

classifying both transformed and non-transformed sensor data into a plurality of quality categories according to the statistical degree to which the data matches the expected value or range of values;

combining statistical evaluations of both transformed data and non-transformed data into a single unified quality assessment value that is indicative of the comprehensive quality of a particular turbine operational event; and

continuously tracking and updating said unified quality assessment value over time and identifying when a deviation in said assessment value exceeds a pre-determined threshold or range as indicative of a potential system or component failure.