초록 ▼

A sensor error compensation system and method is provided that facilitates improved sensor error detection and compensation. The sensor error compensation system includes an expected value generator and a sensor fault detector. The expected value generator receives sensor data from a plurality of se

A sensor error compensation system and method is provided that facilitates improved sensor error detection and compensation. The sensor error compensation system includes an expected value generator and a sensor fault detector. The expected value generator receives sensor data from a plurality of sensors in a turbine engine. From the sensor data, the expected value generator generates an expected sensor value for a sensor under test. The expected sensor value is passed to the sensor fault detector. The sensor fault detector compares the expected sensor value to a received sensor value to determine if a sensor error has occurred in the sensor under test. If an error has occurred, the error can be compensated for by generating a replacement sensor value to substitute for erroneous received sensor value.

대표청구항 ▼

1. A sensor error compensation system for compensating sensor errors in a turbine engine, the sensor error compensation system comprising: an expected value generator, the expected value generator receiving sensor data from a plurality of turbine engine sensors and generating an expected sensor valu

1. A sensor error compensation system for compensating sensor errors in a turbine engine, the sensor error compensation system comprising: an expected value generator, the expected value generator receiving sensor data from a plurality of turbine engine sensors and generating an expected sensor value for a sensor under test; and a sensor fault detector, the sensor fault detector comparing the sensor expected value to a received sensor value for the sensor under test to determine if a sensor error has occurred in the sensor under test. 2. The system of claim 1 wherein the sensor error compensation system further generates a replacement sensor value to substitute for the received sensor value. 3. The system of claim 2 wherein the replacement sensor value is generated from the expected sensor value. 4. The system of claim 3 wherein the replacement sensor value is generated from the expected sensor value by the expected value generator generating a new expected sensor value based on the expected sensor value as an input. 5. The system of claim 1 wherein the expected value generator comprises an auto-associative neural network, and wherein the sensor under test comprises one of the plurality of turbine engine sensors. 6. The system of claim 1 wherein the expected value generator comprises hetero-associative neural network, and wherein the sensor under test comprises an additional turbine engine sensor. 7. The system of claim 1 wherein the expected value generator generates a new expected value using the expected value as an input to isolate the sensor under test from the plurality of turbine engine sensors. 8. The system of claim 1 wherein the expected value generator comprises an auto-associative neural network, and wherein the auto-associative neural network includes a mapping layer, a bottleneck layer and a demapping layer. 9. The system of claim 8 wherein the auto-associative neural network is first trained for association using nominal sensor values, the demapping layer is then fixed and the neural network is trained with faulty sensor values. 10. The system of claim 1 wherein the expected value generator includes a model based on groupings of the plurality of turbine engine sensors into related groups. 11. The system of claim 10 wherein the groupings of the plurality of turbine engine sensors is based upon a time dependent correlation. 12. A method of compensating for sensor errors in a turbine engine, the method comprising the steps of: a) receiving a plurality of sensor values from a plurality of sensors; b) generating an expected sensor value from the plurality of sensor values; c) comparing the expected sensor value to a received sensor value from a sensor under test; and d) replacing the received sensor value with a replacement sensor value if a sensor error in the sensor under test has occurred. 13. The method of claim 12 wherein the step of generating an expected sensor value from a plurality of sensor values comprises generated the expected sensor value with an auto-associative neural network, and wherein the sensor under test comprises one of the plurality of turbine engine sensors. 14. The method of claim 13 wherein the step of generating an expected sensor value from a plurality of sensor values comprises inputting the plurality of sensor values into the auto-associative neural network and processing the plurality of sensor values through a mapping layer, bottleneck layer and demapping layer in the auto-associative neural network. 15. The method of claim 12 wherein the step of generating an expected sensor value from a plurality of sensor values comprises generated the expected sensor value with hetero-associative neural network. 16. The method of claim 12 further comprising the step of isolating the sensor under test from the plurality of turbine engine sensors. 17. The method of claim 16 wherein the step of isolating the sensor under test from the plurality of turbine engine sensors comprises generating a new expected value using the expected value as an input. 18. The method of claim 12 wherein the step of comparing the expected sensor value to a received sensor value from a sensor under test comprises determining a residual difference between the expected sensor value and the received sensor value. 19. The method of claim 12 wherein the step of replacing the received sensor value with a replacement sensor value comprises the step of generating the replacement sensor value from the expected sensor value. 20. The method of claim 19 wherein the step of generating the replacement sensor value from the expected sensor value comprises generating a new expected sensor value using the expected sensor value as an input. 21. An apparatus comprising: a) a processor; b) a memory coupled to the processor; c) a sensor error compensation program residing in the memory and being executed by the processor, the sensor error compensation program including: an expected value generator, the expected value generator receiving sensor data from a plurality of turbine engine sensors and generating an expected sensor value for a sensor under test; anda sensor fault detector, the sensor fault detector comparing the sensor expected value to a received sensor value for the sensor under test to determine if a sensor error has occurred in the sensor under test. 22. The apparatus of claim 20 wherein the sensor error compensation program further generates a replacement sensor value to substitute for the received sensor value. 23. The apparatus of claim 22 wherein the replacement sensor value is generated from the expected sensor value. 24. The apparatus of claim 23 wherein the replacement sensor value is generated from the expected sensor value by the expected value generator generating a new expected sensor value based on the expected sensor value as an input. 25. The apparatus of claim 20 wherein the expected value generator comprises an auto-associative neural network, and wherein the sensor under test comprises one of the plurality of turbine engine sensors. 26. The apparatus of claim 20 wherein the expected value generator comprises hetero-associative neural network, and wherein the sensor under test comprises an additional turbine engine sensor. 27. The apparatus of claim 20 wherein the expected value generator generates a new expected value using the expected value as an input to isolate the sensor under test from the plurality of turbine engine sensors. 28. The apparatus of claim 20 wherein the expected value generator comprises an auto-associative neural network, and wherein the auto-associative neural network includes a mapping layer, a bottleneck layer and a demapping layer. 29. The apparatus of claim 28 wherein the auto-associative neural network is first trained for association using nominal sensor values, the demapping layer is then fixed and the neural network is trained with faulty sensor values. 30. A program product comprising: a) a sensor error compensation program, the sensor error compensation program including: an expected value generator, the expected value generator receiving sensor data from a plurality of turbine engine sensors and generating an expected sensor value for a sensor under test; anda sensor fault detector, the sensor fault detector comparing the sensor expected value to a received sensor value for the sensor under test to determine if a sensor error has occurred in the sensor under test; and b) signal bearing media bearing said program. 31. The program product of claim 30 wherein the signal bearing media comprises recordable media. 32. The program product of claim 30 wherein the signal bearing media comprises transmission media. 33. The program product of claim 30 wherein the sensor error compensation program further generates a replacement sensor value to substitute for the received sensor value. 34. The program product of claim 33 wherein the replacement sensor value is generated from the expected sensor value. 35. The program product of claim 34 wherein the replacement sensor value is generated from the expected sensor value by the expected value generator generating a new expected sensor value based on the expected sensor value as an input. 36. The program product of claim 30 wherein the expected value generator comprises an auto-associative neural network, and wherein the sensor under test comprises one of the plurality of turbine engine sensors. 37. The program product of claim 30 wherein the expected value generator comprises hetero-associative neural network, and wherein the sensor under test comprises an additional turbine engine sensor. 38. The program product of claim 30 wherein the expected value generator generates a new expected value using the expected value as an input to isolate the sensor under test from the plurality of turbine engine sensors. 39. The program product of claim 30 wherein the expected value generator comprises an auto-associative neural network, and wherein the auto-associative neural network includes a mapping layer, a bottleneck layer and a demapping layer. 40. The program product of claim 39 wherein the auto-associative neural network is first trained for association using nominal sensor values, the demapping layer is then fixed and the neural network is trained with faulty sensor values.