Gas turbine combustion profile monitoring

• US 9,791,351 B2
• Filed: 02/06/2015
• Issued: 10/17/2017
• Est. Priority Date: 02/06/2015
• Status: Active Grant

• Current Assignee: GE Infrastructure Technology, LLC (GE Vernova, Inc.)
• Original Assignee: General Electric Company
• Inventors: Miller, Karen Warren, Yu, Lijie, Iasillo, Robert
• Primary Examiner(s): Larkin, Daniel S
• Assistant Examiner(s): Megna Fuentes, Anthony W
• Application Number:
  US14/616,105
• Publication Number:
  US 20160231199A1
• Time in Patent Office: 984 Days
• Field of Search: 7311201
• CPC Class Codes: F02C 3/04 having a turbine driving a compressor power transmission arrangements F02C7/36; control of working fluid flow F02C9/16
• CPC Class Codes: F02C 9/00 Controlling gas-turbine plants; Controlling fuel supply in air- breathing jet-propulsion plants controlling air intakes F02C7/057; controlling turbines F01D; controlling compressors F04D27/00; controlling in general G05
• CPC Class Codes: F05D 2260/80 Diagnostics
• CPC Class Codes: F05D 2260/83 Testing, e.g. methods, components or tools therefor
• CPC Class Codes: F05D 2270/0831 indirectly, at the exhaust
• CPC Class Codes: G01M 15/14 Testing gas-turbine engines or jet-propulsion engines

Abstract

Systems and methods for gas turbine combustion profile monitoring are disclosed. In one embodiment, a method for detecting an anomaly in a combustion section of a gas turbine is disclosed. The method includes receiving, from a plurality of thermal sensors disposed around an exhaust section of a gas turbine, exhaust profile data of the gas turbine. The method further analyzes the exhaust profile data to calculate statistical features associated with a peak-trough pattern. The method further determines, using a machine learning algorithm, that the statistical features are abnormal. In response to the determination, the method processes the exhaust profile data for a predetermined period of time and reports an anomaly in a combustion section of the gas turbine if the statistical features remain abnormal for the predetermined period of time.

Claims

1. A system for detecting an anomaly in a combustion section of a gas turbine, the system comprising:

a plurality of combustors associated with a combustion section;

a plurality of thermal sensors disposed at an exhaust section of a gas turbine, wherein the thermal sensors are configured to provide exhaust profile data of the gas turbine;

a processing circuit communicatively coupled to a memory, the memory storing instructions which when executed by the processing circuit perform operations comprising;

receiving, from the plurality of thermal sensors, the exhaust profile data of the gas turbine;

analyzing the exhaust profile data to calculate statistical features associated with a peak-trough pattern;

determining, using an algorithm or model, that the statistical features are abnormal; and

in response to the determination, continuing processing of the exhaust profile data for a predetermined period of time; and

reporting an anomaly in the combustion section of the gas turbine when the statistical features remain abnormal for the predetermined period of time.

2. The system of claim 1, wherein the algorithm or model is based on historical exhaust profile data, the historical exhaust profile data including normal data samples and false data samples.

3. The system of claim 1, wherein the thermal sensors are positioned radially around an exhaust component associated with the turbine.

4. The system of claim 1, wherein the sensors are disposed in an evenly spaced array.

5. The system of claim 1, wherein the exhaust profile data includes a plurality of peak and trough couples, each peak and trough couple of the plurality of peaks and troughs corresponding to at least one combustor of the plurality of the combustors.

6. The system of claim 5, wherein the analyzing includes evaluating each peak and trough couple relative to an expected peak-trough pattern.

7. The system of claim 5, wherein analyzing includes:

identifying, based on the exhaust profile data, a peak and trough couple associated with each thermal sensor; and

calculating the statistical features for each peak and trough couple.

8. The system of claim 1, wherein the exhaust profile data includes statistical features associated with the plurality of thermal sensors, the statistical features including at least one of the following:

a minimum peak temperature, a maximum peak temperature, a minimum trough temperature, a maximum trough temperature, a minimum peak-to-trough delta, and a maximum peak-to-trough delta.

9. The system of claim 1, wherein the determining that the statistical features are abnormal includes creating a feature vector based at least on the statistical features and processing the feature vector through a classification model.

10. The system of claim 1, further comprising prior to the analyzing:

determining a quality of the exhaust profile data versus a predetermined quality level; and

in response to the determining that the quality is below a predetermined quality level, adjusting the exhaust profile data.

11. The system of claim 1, wherein the analyzing is performed after the gas turbine is operating above a predetermined load.

12. The system of claim 1, further comprising issuing an alarm based at least in part on the detecting of the anomaly in the combustion section of the gas turbine.

13. The system of claim 12, wherein the alarm triggers at least one of the following:

a further evaluation and a responsive action.

14. A method for detecting an anomaly in a combustion section of a gas turbine, the method comprising:

receiving, from a plurality of thermal sensors disposed at an exhaust section of a gas turbine, exhaust profile data of the gas turbine;

analyzing the exhaust profile data to calculate statistical features associated with a peak-trough pattern;

determining, using an algorithm or model, that the statistical features are abnormal; and

in response to the determination, continuing processing of the exhaust profile data for a predetermined period of time; and

reporting an anomaly in a combustion section of the gas turbine when the statistical features remain abnormal for the predetermined period of time.

15. The method of claim 14, wherein:

the algorithm or model is based on historical exhaust profile data, the historical exhaust profile data including normal data samples and false data samples.

16. The method of claim 14, wherein the thermal sensors are:

positioned radially around an exhaust component associated with the gas turbine and disposed in an evenly spaced array.

17. The method of claim 14, wherein the exhaust profile data includes a plurality of peak and trough couples, each peak and trough couple of the plurality of peaks and troughs corresponding to at least one combustor of the plurality of combustors.

18. The method of claim 17, wherein the calculating the statistical features includes:

identifying, based on the exhaust profile data, a peak and trough couple associated with each thermal sensor; and

calculating statistical features for each peak and trough couple.

19. The method of claim 14, wherein the determining that the statistical features are abnormal includes creating a feature vector based at least on the statistical features and processing the feature vector through a classification model.

20. A non-transitory computer-readable medium having stored instructions, which when executed by at least one processor, perform operations comprising:

receiving, from a plurality of thermal sensors disposed at an exhaust section of a gas turbine, exhaust profile data of the gas turbine;

analyzing the exhaust profile data to calculate statistical features associated with a peak-trough pattern;

determining, using an algorithm or model, that the statistical features are abnormal; and

in response to the determination, continuing processing of the exhaust profile data for a predetermined period of time; and

reporting the anomaly in a combustion section of a gas turbine when the statistical features remain abnormal for the predetermined period of time.