Methods for monitoring and controlling boiler flames

• US 7,353,140 B2
• Filed: 04/26/2005
• Issued: 04/01/2008
• Est. Priority Date: 11/14/2001
• Status: Expired due to Term

• Current Assignee: Electric Power Research Institute Incorporated
• Original Assignee: Electric Power Research Institute Incorporated
• Inventors: Daw, Charles Stuart, Flynn, Thomas J., Finney, Charles E. A., Fuller, Timothy A.
• Primary Examiner(s): Cosimano; Edward R
• Application Number:
  US11/115,625
• Publication Number:
  US 20060015298A1
• Time in Patent Office: 1,071 Days
• Field of Search: 738/659, 324/500, 340/3.1, 340/3.43, 340/500, 340/635, 382/100, 382/155, 382/159, 431/18, 431/24, 700/90, 700/108, 702/33, 702/34, 702/35, 702/127, 702/182, 702/183, 702/184, 702/185, 702/186, 702/187
• US Class Current: 702/182
• CPC Class Codes: F23M 11/045 by observing the flame
• CPC Class Codes: F23N 2223/44 Optimum control
• CPC Class Codes: F23N 2229/04 sensitive to the colour of flames
• CPC Class Codes: F23N 2229/08 detecting flame flicker
• CPC Class Codes: F23N 5/08 using light-sensitive elements
• CPC Class Codes: F23N 5/16 using noise-sensitive detectors

Abstract

The current invention provides a method and apparatus, which uses symbol sequence techniques, temporal irreversibility, and/or cluster analysis to monitor the operating state of individual burner flames on a appropriate time scale. Both the method and apparatus of the present invention may be used optimize the performance of burner flames.

Claims

1. A method of classifying the flame state of a burner flame, comprising:

obtaining a series of data over a predetermined period of time for a burner flame;

comparing said series of data for said burner flame to a library of clusters, wherein each of said clusters in said library is categorized as a particular burner flame state;

identifying one of said clusters in said library that is a statistical best match to said series of data for said burner flame;

classifying the flame state of said burner flame based upon said one of said clusters that is said statistical best match; and

outputting the classification of the flame state.

2. The method of claim 1, further comprising constructing said library of clusters based upon previously measured flame states.

3. The method of claim 1, wherein said comparing comprises:

computing at least one statistic for said series of data for said burner flame;

computing, for each of said clusters in said library, a corresponding cluster mean of said at least one statistic; and

computing, for each of said clusters in said library, a corresponding normalized statistic based upon said at least one statistic for said series of data for said burner flame and said corresponding cluster mean to produce a group of corresponding normalized statistics, each providing a degree of statistical match between said series of data for said burner flame and each of said clusters in said library.

4. The method of claim 3, wherein said computing said corresponding normalized statistic further comprises computing, for each of said clusters in said library, said corresponding normalized statistic based on a standard deviation for said at least one statistic.

5. The method of claim 3, wherein said identifying further comprises identifying a smallest one of said corresponding normalized statistics.

6. The method of claim 3, wherein said at least one statistic comprises a form of a scalar and a vector.

7. The method of claim 3, wherein said at least one statistic comprises skewness, kurtosis, or both.

8. The method of claim 3, wherein said at least on statistic comprises a symbol histogram, a time asymmetry function, or both.

9. The method of claim 8, wherein said time asymmetry function comprises a low-passband time asymmetry function or a high-passband time asymmetry function.

10. A method of evaluating the flame state of a burner flame, comprising:

obtaining a series of data over a predetermined period of time for a burner flame;

comparing said series of data for said burner flame to a library of clusters, wherein each of said clusters in said library is categorized as a particular burner flame state;

identifying one of said clusters in said library that is a statistical best match to said series of data for said burner flame;

classifying the flame state of said burner flame based upon said one of said clusters that is said statistical best match;

identifying a root cause of said flame state of said burner flame; and

outputting the identification of the root cause.

11. The method of claim 10, further comprising constructing said library of clusters based upon previously measured flame states.

12. The method of claim 10, wherein said comparing comprises:

computing at least one statistic for said series of data for said burner flame;

computing, for each of said clusters in said library, a corresponding cluster mean of said at least one statistic; and

computing, for each of said clusters in said library, a corresponding normalized statistic based upon said at least one statistic for said series of data for said burner flame and said corresponding cluster mean to produce a group of corresponding normalized statistics, each providing a degree of statistical match between said series of data for said burner flame and each of said clusters in said library.

13. The method of claim 12, wherein said computing said corresponding normalized statistic further comprises computing, for each of said clusters in said library, said corresponding normalized statistic based on a standard deviation for said at least one statistic.

14. The method of claim 12, wherein said identifying further comprises identifying a smallest one of said corresponding normalized statistics.

15. The method of claim 12, wherein said at least one statistic comprises a form of a scalar and a vector.

16. The method of claim 12, wherein said at least one statistic comprises skewness, kurtosis, or both.

17. The method of claim 12, wherein said at least on statistic comprises a symbol histogram, a time asymmetry function, or both.

18. The method of claim 17, wherein said time asymmetry function comprises a low-passband time asymmetry function or a high-passband time asymmetry function.

19. A method of optimizing the flame state of a burner flame, comprising:

obtaining a series of data over a predetermined period of time for a burner flame;

comparing said series of data for said burner flame to a library of clusters, wherein each of said clusters in said library is categorized as a particular burner flame state;

identifying one of said clusters in said library that is a statistical best match to said series of data for said burner flame;

classifying the flame state of said burner flame based upon said one of said clusters that is said statistical best match;

identifying a root cause of any non-optimal flame state of said burner flame and an effect of said root cause; and

reducing said effect of said root cause on said burner flame.

20. The method of claim 19, further comprising constructing said library of clusters based upon previously measured flame states.

21. The method of claim 19, wherein said comparing comprises:

computing at least one statistic for said series of data for said burner flame;

computing, for each of said clusters in said library, a corresponding cluster mean of said at least one statistic; and

computing, for each of said clusters in said library, a corresponding normalized statistic based upon said at least one statistic for said series of data for said burner flame and said corresponding cluster mean to produce a group of corresponding normalized statistics, each providing a degree of statistical match between said series of data for said burner flame and each of said clusters in said library.

22. The method of claim 21, wherein said computing said corresponding normalized statistic further comprises computing, for each of said clusters in said library, said corresponding normalized statistic based on a standard deviation for said at least one statistic.

23. The method of claim 21, wherein said identifying further comprises identifying a smallest one of said corresponding normalized statistics.

24. The method of claim 21, wherein said at least one statistic comprises a form of a scalar and a vector.

25. The method of claim 21, wherein said at least one statistic comprises skewness, kurtosis, or both.

26. The method of claim 21, wherein said at least on statistic comprises a symbol histogram, a time asymmetry function, or both.

27. The method of claim 26, wherein said time asymmetry function comprises a low-passband time asymmetry function or a high-passband time asymmetry function.