초록 ▼

A method for detecting a fault in a physical system uses a model of the physical system and calculates estimated dependent variables or conditions for the system using substantially only independent variables that are measured from the system using hardware redundancy or selected based on their bett

A method for detecting a fault in a physical system uses a model of the physical system and calculates estimated dependent variables or conditions for the system using substantially only independent variables that are measured from the system using hardware redundancy or selected based on their better measurement reliability. An example of hardware redundancy is to measure an independent variable using two or more sensors rather than one. The estimated dependent variables are compared to the corresponding measured dependent variable conditions to calculate residuals, which are then analyzed using appropriate fault detection techniques. The method is especially effective relative to prior fault detection method when used to detect anomalies or unknown fault states of the system.

대표청구항 ▼

1. A method for detecting a fault in a system, comprising:obtaining a measurement for each of a plurality of variables corresponding to the operational state of the system; ranking the variables by the reliability of the measurement for each of the plurality of variables to provide a ranked list of

1. A method for detecting a fault in a system, comprising:obtaining a measurement for each of a plurality of variables corresponding to the operational state of the system; ranking the variables by the reliability of the measurement for each of the plurality of variables to provide a ranked list of the variables; selecting a first subset of the most reliable variables from the ranked list to provide a set of independent variables; and calculating expected system dependent variables using the set of independent variables to detect the fault. 2. The method of claim 1 wherein calculating expected system dependent variables comprises:developing a model corresponding to the system; and formulating the model to calculate the expected system dependent variables based on the set of independent variables. 3. The method of claim 2 wherein the model comprises physical model corresponding to the system.4. The method of claim 3 wherein the physical model comprises model update scheme.5. The method of claim 4 wherein the model update scheme comprises a correction factor based on data collected from operation of the system.6. The method of claim 1 wherein at least one of the set of independent variables is measured using hardware redundancy.7. The method of claim 6 wherein the hardware redundancy comprises measuring the at least one of the set of independent variables using at least two hardware measuring devices.8. The method of claim 7 wherein the measuring devices are sensors.9. The method of claim 1 wherein the fault is an anomaly.10. A computer-readable medium comprising a computer program for operating a computer system to detect a fault in a physical system according to the method of claim 1.11. The method of claim 1 further comprising:selecting a second subset of variables from the ranked list to provide a set of measured dependent variables; and comparing the expected system dependent variables to the set of measured dependent variables. 12. The method of claim 11 wherein the comparing comprises residual analysis.13. The method of claim 1 wherein the system is selected from the group consisting of an aircraft engine, a rocket propulsion system, and an aerospace vehicle.14. The method of claim 1 wherein measuring the plurality of variables comprises sensing operating conditions using at least one sensor.15. The method of claim 1 wherein the reliability of the measurement is based on the confidence of accurately obtaining the measurement for each of the plurality of variables.16. The method of claim 1 wherein the expected system dependent variables are calculated substantially only using the set of independent variables.17. A computer system for detecting an anomaly in a physical system, comprising:means for providing a model of the physical system; means for receiving a plurality of sensor measurements from the physical system, wherein at least one sensor measurement is used as an independent variable and at least one sensor measurement is used as an actual sensor measurement; means for processing the independent variable through the model of the physical system to generate an estimated variable as a function of the independent variable; and mean for comparing the estimated variable and the actual sensor measurement to determine an anomaly in the physical system. 18. The computer system of claim 17, further including a plurality of sensors coupled to the physical system for providing the plurality of sensor measurements.19. The computer system of claim 18, wherein the plurality of sensors measure physical states of the physical system.20. The computer system of claim 19, wherein the plurality of sensors include redundant sensors for a physical s ate of the physical system.21. The computer system of claim 17, wherein a residual between the estimated variable and the actual sensor measurement is determined.22. The computer system of claim 21, wherein the residual is compared to a threshold to determine the anomaly in the physical system.23. The computer system of claim 21, wherein the residual is processed through a classification procedure to determine the anomaly in the physical system.24. The computer system of claim 17, wherein the model of the physical system is given in the form of y=f(x)*n(x, t), where y is the estimated variable, x is the independent variable, t is time, f(x) is a physical response, and n(x, t) is a correction factor.25. The computer system of claim 17, wherein the physical system is an engine.26. An apparatus for detecting an anomaly in a system, comprising:a plurality of sensors coupled to the system for providing sensor measurements, wherein a first sensor measurement represents an independent variable and a second sensor measurement represents an actual sensor measurement; and a computational system providing a model of the system, the computational system including, a) means for processing the independent variable through the model of the system to generate an estimated variable as a function of the independent variable, and b) means for comparing the estimated variable and the actual sensor measurement to determine an anomaly in the system. 27. The apparatus of claim 26, wherein a residual between the estimated variable and the actual sensor measurement is determined.28. The apparatus of claim 27, wherein the residual is compared to a threshold to determine the anomaly in the system.29. The apparatus of claim 27, wherein the residual is processed through a classification procedure to determine the anomaly in the system.30. The apparatus of claim 26, wherein the plurality of sensors measure physical states of the system.31. The apparatus of claim 30, wherein the plurality of sensors include redundant sensors for a physical state of the system.32. The apparatus of claim 26, wherein the model of the system is given in the form of y=f(X)*n(x, t), where y is the estimated variable, x is the independent variable, t is time, if (x) is a physical response, and n(x, t) is a correction factor.33. The apparatus of claim 26, wherein the system is an engine.34. A system analysis tool for detecting an anomaly in a system, comprising;a plurality of sensors coupled to the system for providing sensor measurements, wherein a first sensor measurement is used an independent variable and a second sensor measurement is used an actual sensor measurement; and a model of the system, the model including, (a) means for generating an estimated variable as a function of the independent variable, and (b) means for comparing the estimated variable and the actual sensor measurement to determine an anomaly in the system. 35. The system analysis tool of claim 34, wherein a residual between the estimated variable and the actual sensor measurement is determined.36. The system analysis tool of claim 35, wherein the residual is compared to a threshold to determine the anomaly in the system.37. The system analysis tool of claim 35, wherein the residual is processed through a classification procedure to determine the anomaly in the system.38. The system analysis tool of claim 34, wherein the plurality of sensors measure physical states of the system.39. The system analysis tool of claim 38, wherein the plurality of sensors include redundant sensors for a physical state of the system.40. The system analysis tool of claim 34, wherein the model of the system is given in the form of y=f(x)*n(x, t), where y is the estimated variable, x is the independent variable, t is time, f(x) is a physical response, and n(x, t) is a correction factor.41. The system analysis tool of claim 34, wherein the system is an engine.42. A method for detecting an anomaly in a physical system, comprising:providing a model of the physical system; receiving a plurality of sensor measurements from the physical system, wherein a first sensor measurement is used as an independent variable and a second sensor measurement is used as an actual sensor measurement, processing the independent variable through the model of the physical system to generate an estimated variable as a function of the independent variable; and comparing the estimated variable and the actual sensor measurement to determine an anomaly in the physical system. 43. The method of claim 42, further including determining a residual between the estimated variable and the actual sensor measurement.44. The method of claim 43, wherein the residual is compared to a threshold to determine the anomaly in the physical system.45. The method of claim 43, wherein the residual is processed through a classification procedure to determine the anomaly in the physical system.46. The method of claim 42, further including providing a plurality of sensors coupled to the physical system for providing the plurality of sensor measurements.47. The method of claim 46, wherein the plurality of sensors measure physical states of the physical system.48. The method of claim 42, wherein the model of the physical system is given in the form of y=f(x)*n(x, t), where y is the estimated variable, x is the independent variable, t is time, f(x) is a physical response, and n(x, t) is a correction factor.