METHOD AND SYSTEM FOR DIAGNOSING FAULTS IN A PARTICULAR DEVICE WITHIN A FLEET OF DEVICES

• US 20080243328A1
• Filed: 04/02/2007
• Published: 10/02/2008
• Est. Priority Date: 04/02/2007
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Yu, Lijie, Cleary, Daniel Joseph, Osborn, Mark David
• Granted Patent: US 7,826,943 B2
• Time in Patent Office: Days
• US Class Current: 701/29
• CPC Class Codes: G05B 23/0232 based on qualitative trend analysis, e.g. system evolution

Abstract

A method for diagnosing faults in a particular device within a fleet of devices is provided. The method comprises receiving performance data related to one or more parameters associated with a fleet of devices and processing the performance data to detect one or more trend shifts in the one or more parameters. The method then comprises detrending the one or more parameters to derive noise-adjusted performance data related to a particular parameter associated with a particular device. The method further comprises generating a fleet-based diagnostic model based on trend patterns and data characteristics associated with the fleet of devices. The fleet-based diagnostic model comprises one or more fuzzy rules defining one or more expected trend shift data ranges for the one or more parameters associated with the fleet of devices. The method then comprises computing one or more scaling factors for the particular parameter associated with the particular device and scaling the one or more of fuzzy rules defined for the one or more parameters in the fleet-based diagnostic model, based on the one or more scaling factors, to generate a personalized diagnostic model for the particular parameter associated with the particular device. The method finally comprises evaluating the personalized diagnostic model against the one or more trend shifts detected for the one or more parameters, to diagnose a fault associated with the particular device.

Claims

1. A method for diagnosing faults in a particular device within a fleet of devices, the method comprising:

receiving performance data related to one or more parameters associated with a fleet of devices;

processing the performance data to detect one or more trend shifts in the one or more parameters;

detrending the one or more parameters to derive noise-adjusted performance data related to a particular parameter associated with a particular device;

generating a fleet-based diagnostic model based on trend patterns and data characteristics associated with the fleet of devices, wherein the fleet-based diagnostic model comprises one or more fuzzy rules defining one or more expected trend shift data ranges for the one or more parameters associated with the fleet of devices;

computing one or more scaling factors for the particular parameter associated with the particular device;

scaling the one or more of fuzzy rules defined for the one or more parameters in the fleet-based diagnostic model, based on the one or more scaling factors, to generate a personalized diagnostic model for the particular parameter associated with the particular device; and

evaluating the personalized diagnostic model against the one or more trend shifts detected for the one or more parameters, to diagnose a fault associated with the particular device.

2. The method of claim 1, wherein each of the devices comprise one of a mechanical system, an electrical system, an electromechanical system or a biological system.

3. The method of claim 1, wherein the devices comprise aircraft engine systems.

4. The method of claim 1, wherein the parameters comprise at least one of engine exhaust gas temperature, engine fuel flow and engine core speed.

5. The method of claim 1, wherein processing the performance data comprises removing outlying data points from the performance data.

6. The method of claim 1, wherein processing the performance data comprises applying a set of piecewise regressions to detect one or more linear trends in one or more data segments in the performance data.

7. The method of claim 1, wherein de-trending the one or more parameters comprises filtering the one or more trend shifts caused by special events and normal deterioration associated with the particular device.

8. The method of claim 1, wherein the one or more scaling factors is calculated as a ratio of the standard deviation of the one or more de-trended parameters associated with the particular device, and a fleet average, wherein the fleet average is derived based on one or more fleet characteristics associated with the fleet of devices.

9. The method of claim 1, further comprising normalizing the one or more scaling factors for the particular parameter, based on one or more parameter threshold values.

10. The method of claim 1, wherein the one or more fuzzy rules defined for the one or more parameters are represented by one or more fuzzy membership functions.

11. The method of claim 10, wherein the one or more fuzzy membership functions are represented by a trapezoid, and wherein the area defined by the trapezoid represents the one or more expected trend shift data ranges defined by the fuzzy rules for the one or more parameters.

12. The method of claim 11, wherein scaling the one or more fuzzy rules to generate the personalized diagnostic model for the particular parameter associated with the particular device comprises shifting the one or more expected trend shift data ranges, for the parameter, based on the scaling factors.

13. A system for diagnosing faults in a particular device within a fleet of devices, the system comprising:

a trend shift detection component, configured to receive performance data related to one or more parameters associated with a fleet of devices and process the performance data to detect one or more trend shifts in the one or more parameters;

a data detrending component configured to detrend the one or more parameters to derive noise-adjusted performance data related to a particular parameter associated with a particular device;

a fleet-based diagnostic model configured to generate trend patterns and data characteristics associated with the fleet of devices, wherein the fleet-based diagnostic model comprises one or more fuzzy rules defining one or more expected trend shift data ranges for the one or more parameters associated with the fleet of devices;

a scaling component configured to compute one or more scaling factors for the particular parameter associated with the particular device;

a personalized diagnostic model component configured to scale the one or more of fuzzy rules defined for the one or more parameters in the fleet-based diagnostic model, based on the one or more scaling factors, to generate a personalized diagnostic model for the particular parameter associated with the particular device; and

a diagnosis component configured to evaluate the personalized diagnostic model against the one or more trend shifts detected for the one or more parameters, to diagnose a fault associated with the particular device.

14. The system of claim 13, wherein the devices comprise aircraft engine systems and wherein the parameters comprise at least one of engine exhaust gas temperature, engine fuel flow and engine core speed.

15. The system of claim 13, wherein the data detrending component is configured to filter the one or more trend shifts caused by special events and normal deterioration associated with the particular device.

16. The system of claim 13, wherein the scaling component is configured to compute the one or more scaling factors as a ratio of the standard deviation of the one or more de-trended parameters associated with the particular device, and a fleet average, wherein the fleet average is derived based on one or more fleet characteristics associated with the fleet of devices.

17. The system of claim 13, wherein the one or more fuzzy rules defined for the one or more parameters are represented by one or more fuzzy membership functions.

18. The system of claim 17, wherein the one or more fuzzy membership functions are represented by a trapezoid, and wherein the area defined by the trapezoid represents the one or more expected trend shift data ranges defined by the fuzzy rules for the one or more parameters.

19. The system of claim 18, wherein the personalized diagnostic model component is configured to generate the personalized diagnostic model by shifting the one or more expected trend shift data ranges, for the parameter, based on the scaling factors.

20. A method for diagnosing faults for a subset of devices in a fleet of devices, the method comprising:

receiving performance data related to one or more parameters associated with a fleet of devices;

processing the performance data to detect one or more trend shifts in the one or more parameters;

detrending the one or more parameters to derive noise-adjusted performance data related to a particular parameter associated with a subset of devices, wherein the subset of devices comprise devices in the fleet of devices having similar data characteristics and similar fault data;

generating a fleet-based diagnostic model based on trend patterns and data characteristics associated with the fleet of devices, wherein the fleet-based diagnostic model comprises one or more fuzzy rules defining one or more expected trend shift data ranges for the one or more parameters associated with the fleet of devices;

computing one or more scaling factors for the particular parameter associated with the subset of devices;

scaling the one or more of fuzzy rules defined for the one or more parameters in the fleet-based diagnostic model, based on the one or more scaling factors to generate a personalized diagnostic model for the particular parameter associated with the subset of devices; and

evaluating the personalized diagnostic model against the one or more trend shifts detected for the one or more parameters, to diagnose a fault associated with the subset of devices.