Method and system for hierarchical fault classification and diagnosis in large systems

• US 7,379,799 B2
• Filed: 06/29/2005
• Issued: 05/27/2008
• Est. Priority Date: 06/29/2005
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Osborn, Mark David, Yu, LiJie, Cleary, Daniel Joseph
• Primary Examiner(s): ARTHUR JEANGLAUDE, GERTRUDE
• Application Number:
  US11/171,961
• Publication Number:
  US 20080039993A1
• Time in Patent Office: 1,063 Days
• Field of Search: 701/2, 701/29, 701/30, 701/33, 701 34- 36, 340/425.5, 340/438, 340/439
• US Class Current: 701/29
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

A method for diagnosing and classifying faults in a system is provided. The method comprises acquiring operational data for at least one of a system, one or more subsystems of the system or one or more components of the one or more subsystems. Then, the method comprises analyzing the operational data using one or more diagnostic models. Each diagnostic model uses the operational data to determine a probability of fault associated with at least one of the one or more components or the one or more subsystems. Finally, the method comprises deriving an overall probability of fault for at least one of the system, the one or more subsystems, or the one or more components using the one or more probabilities of fault determined by the one or more diagnostic models and one or more hierarchical relationships between the subsystems and components of the system.

Claims

1. A method for diagnosing and classifying faults in a system, the method comprising:

acquiring operational data for at least one of a system, one or more subsystems of the system or one or more components of the one or more subsystems;

analyzing the operational data using one or more diagnostic models, wherein each diagnostic model uses the operational data to determine a probability of fault associated with at least one of the one or more components or the one or more subsystems; and

deriving an overall probability of fault for at least one of the system, the one or more subsystems, or the one or more components using the one or more probabilities of fault determined by the one or more diagnostic models and one or more hierarchical relationships between the subsystems and components of the system.

2. The method of claim 1, wherein the system is an aircraft engine system.

3. The method of claim 1, wherein the step of deriving utilizes a probabilistic analysis comprising leaf nodes and parent nodes.

4. The method of claim 1, wherein the probability of fault derived by the one or more diagnostic models is propagated through a hierarchical fault network.

5. The method of claim 4, wherein the hierarchical fault network is a Bayesian Belief Network (BBN).

6. The method of claim 1, wherein the diagnostic models include at least one of a physics-based diagnostic model, an experience-based diagnostic model, or a regression-based model.

7. The method of claim 6, wherein the physics based diagnostic model generates a probability of fault based on a first principal system model.

8. The method of claim 6, wherein the experience-based diagnostic model generates a probability of fault based on an expert opinion evaluation of the subsystems.

9. The method of claim 1, wherein acquiring the operational data comprises collecting the operational data from a plurality of sensors coupled to at least one or more of the subsystems or the one or more components.

10. The method of claim 1, wherein acquiring the operational data comprises collecting the operational data from a data storage device.

11. A processor-based system, comprising:

an input device configured to receive operational data for at least one of a system, one or more subsystems of the system or one or more components of the one or more subsystems;

a processor configured to execute routines to analyze the operational data using one or more diagnostic models, wherein each diagnostic model uses the operational data to determine a probability of fault associated with at least one of the one or more components or the one or more subsystems and derive an overall probability of fault for at least one of the system, the one or more subsystems, or the one or more components using the one or more probabilities of fault determined by the one or more diagnostic models and one or more hierarchical relationships between the subsystems and components of the system;

a storage device coupled to the processor for storing the results of the analysis of the operational data by the processor; and

an output device coupled to the processor for displaying the results of the analysis.

12. A computer readable media, comprising:

routines adapted to acquire operational data for at least one of a system, one or more subsystems of the system or one or more components of the one or more subsystems;

routines adapted to analyze the operational data using one or more diagnostic models, wherein each diagnostic model uses the operational data to determine a probability of fault associated with at least one of the one or more components or the one or more subsystems; and

routines adapted to derive an overall probability of fault for at least one of the system, the one or more subsystems, or the one or more components using the one or more probabilities of fault determined by the one or more diagnostic models and one or more hierarchical relationships between the subsystems and components of the system.

13. A method, comprising:

acquiring operational data for two or more logical groupings of subsystems, wherein the two or more logical groupings comprise one or more common components;

analyzing the operational data using one or more diagnostic models, wherein each diagnostic model uses the operational data to determine two or more probabilities of fault for each of the one or more common components using the operational data; and

deriving an overall probability of fault for each of the one or more common components based on the two or more probabilities of fault and on a hierarchical relationship between or within the two or more logical groupings of subsystems.

14. The method of claim 13, wherein the step of deriving utilizes a probabilistic analysis comprising leaf nodes and parent nodes.

15. The method of claim 13, wherein the probabilities of fault derived by the one or more diagnostic models is propagated through a hierarchical fault network.

16. The method of claim 15, wherein the hierarchical fault network is a Bayesian Belief Network (BBN).

17. The method of claim 13, wherein the diagnostic models include at least one of a physics-based diagnostic model, an experience-based diagnostic model, or a regression-based model.

18. The method of claim 13, wherein acquiring the operational data comprises collecting the operational data from a plurality of sensors coupled to at least one or more of the common components.

19. A processor-based system, comprising:

an input device configured to receive operational data for two or more logical groupings of subsystems, wherein the two or more logical groupings comprise one or more common components;

a processor configured to execute routines to analyze the operational data using one or more diagnostic models, wherein each diagnostic model uses the operational data to determine two or more probabilities of fault for each of the one or more common components using the operational data and derive an overall probability of fault for each of the one or more common components based on the two or more probabilities of fault and on a hierarchical relationship between or within the two or more logical groupings of subsystems;

a storage device coupled to the processor for storing the results of the analysis of the operational data by the processor; and

an output device coupled to the processor for displaying the results of the analysis.

20. A computer readable media, comprising:

routines adapted to acquire operational data for two or more logical groupings of subsystems, wherein each logical grouping has one or more common components;

routines adapted to analyze the operational data using one or more diagnostic models, wherein each diagnostic model uses the operational data to determine two or more probabilities of fault for each of the one or more common components using the operational data; and

routines adapted to derive an overall probability of fault for each of the one or more common components based on the two or more probabilities of fault and on a hierarchical relationship between or within the two or more logical groupings of subsystems.