Thermocouple failure detection in power generation turbines
First Claim
Patent Images
1. A method of thermocouple failure detection in power generation turbines comprising:
- creating redundancy estimates from temperature readings generated by said thermocouples;
predicting an expected value from each temperature reading;
comparing said temperature readings and said redundancy estimates with said expected value;
fusing said redundancy estimates and said expected values to generate a fused thermocouple value;
and generating a thermocouple confidence by comparing said fused thermocouple value and said temperature readings;
wherein said predicting of expected values is completed by creating a history of the process.
1 Assignment
0 Petitions
Accused Products
Abstract
-
Citations
39 Claims
-
1. A method of thermocouple failure detection in power generation turbines comprising:
-
creating redundancy estimates from temperature readings generated by said thermocouples;
predicting an expected value from each temperature reading;
comparing said temperature readings and said redundancy estimates with said expected value;
fusing said redundancy estimates and said expected values to generate a fused thermocouple value;
and generating a thermocouple confidence by comparing said fused thermocouple value and said temperature readings;
wherein said predicting of expected values is completed by creating a history of the process. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18)
x(k+1)=x(k)+u(k)+w(k) where u(k) is an unknown input to be estimated, w(k) is noise, and x(k) is the temperature detected by a respective thermocouple.
-
-
11. A method in accordance with claim 10 wherein x(k) and w(k) are independent, random and Gaussian distributed.
-
12. A method in accordance with claim 10, wherein u(k) is estimated using a polynomial technique.
-
13. A method in accordance with claim 10, wherein u(k) is estimated using a neural network.
-
14. A method in accordance with claim 10, wherein u(k) is estimated using a fuzzy logic technique.
-
15. A method in accordance with claim 1, wherein comparing said temperature readings and said redundancy estimates with said expected value is completed using a validation gate.
-
16. A method in accordance with claim 15, wherein said validation gate is defined by a 3 sigma area.
-
17. A method in accordance with claim 1, wherein said thermocouple confidence is calculated for each thermocouple within each set of thermocouples that a respective thermocouple belongs.
-
18. A method in accordance with claim 17, wherein a failed thermocouple is detected if the thermocouple confidence for a respective thermocouple is zero in each set of thermocouples that a respective thermocouple belongs.
-
19. A thermocouple failure detection system (10) for a power generation turbine assembly (12) comprising:
-
a plurality of thermocouples (14) positioned to monitor temperature within said turbine assembly (12); and
a computer (16) coupled to said plurality of thermocouples (14);
wherein temperature signals generated by respective thermocouples (14) are provided to said computer (16) and utilized within an algorithm that is embedded within said computer to generate a thermocouple confidence value, which value provides an indication of thermocouple operation or alternatively thermocouple failure;
wherein said algorithm is performed in a language selected from the group consisting of C, C++, JAVA, basic, Visual Basic, MATLAB, and Fortran. - View Dependent Claims (20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39)
creating redundancy estimates from temperature readings generated by said thermocouples;
predicting an expected value from each temperature reading;
comparing said temperature readings and said redundancy estimates with said expected value;
fusing said redundancy estimates and said expected values to generate a fused thermocouple value;
and generating a thermocouple confidence by comparing said fused thermocouple value and said temperature readings.
-
-
22. A system in accordance with claim 21, wherein said redundancy estimates are created by physical redundancy using multiple thermocouples to monitor the same temperature.
-
23. A system in accordance with claim 21, wherein said redundancy estimates are created by functional redundancy using measurements from non-redundant thermocouples to estimate temperature.
-
24. A system in accordance with claim 23, wherein said functional redundancy is represented as xi=ƒ
- (y1, - - - yi−
1, yi+1, - - - yn) where y1, y2 - - - yn are measurements from n thermocouples measuring temperatures at the point each respective thermocouple is located, xi.
- (y1, - - - yi−
-
25. A system in accordance with claim 24, wherein said function is solved using nonlinear regression.
-
26. A system in accordance with claim 24, wherein said function is solved using a neural network.
-
27. A system in accordance with claim 24, wherein said function is solved using an analytical first principal model.
-
28. A system in accordance with claim 23, wherein said functional redundancy estimates are created by monitoring temperature signals at two adjoining thermocouples, one on either side of an at-issue thermocouple.
-
29. A system in accordance with claim 21, wherein said predicting of expected values in completed by creating a history of the process.
-
30. A system in accordance with claim 29, wherein said history is created by building an adaptive time-series model incorporated in a state space form for predicting variables at next sampling time.
-
31. A system in accordance with claim 30, wherein the state of the variable of said next sampling period is:
x(k+1)=x(k)+u(k)+w(k) where u(k) is an unknown input to be estimated, w(k) is noise, and x(k) is the temperature detected by a respective thermocouple.
-
32. A system in accordance with claim 31 wherein x(k) and w(k) are independent, random and Gaussian distributed.
-
33. A system in accordance with claim 31, wherein u(k) is estimated using a polynomial technique.
-
34. A system in accordance with claim 31, wherein u(k) is estimated using a neural network.
-
35. A system in accordance with claim 31, wherein u(k) is estimated using a fuzzy logic technique.
-
36. A system in accordance with claim 21, wherein comparing said temperature readings and said redundancy estimates with said expected value is completed using a validation gate.
-
37. A system in accordance with claim 21, wherein said validation gate is defined by a 3 sigma area.
-
38. A system in accordance with claim 21, wherein said thermocouple confidence is calculated for each thermocouple within each set of thermocouples that a respective thermocouple belongs.
-
39. A system in accordance with claim 38, wherein a failed thermocouple is detected if the thermocouple confidence for a respective thermocouple is zero in each set of thermocouples that a respective thermocouple belongs.
Specification