Methods and apparatus for performing system fault diagnosis

• US 5,099,436 A
• Filed: 11/03/1988
• Issued: 03/24/1992
• Est. Priority Date: 11/03/1988
• Status: Expired due to Term

• Current Assignee: Dio Technology Holdings LLC (Intellectual Ventures LLC)
• Original Assignee: Alliedsignal Inc.
• Inventors: Conway, Timothy J., McCown, Patricia M.
• Primary Examiner(s): Black, Thomas G.
• Application Number:
  US07/266,722
• Time in Patent Office: 1,237 Days
• Field of Search: 364/550, 364/513, 371/26
• US Class Current: 702/82
• CPC Class Codes: G06F 11/2257 using expert systems
• CPC Class Codes: G06F 11/34 Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
• CPC Class Codes: Y10S 706/911 Nonmedical diagnostics

Abstract

A diagnostic tool based on a hybrid knowledge representation of a system under test is disclosed. Data collected from the system during its operation is compared to an event based representation of the system which comprises a plurality of predefined events. An event is recognized when the collected data matches the event'"'"'s critical parameter. The recognized event is analyzed and an associated set of ambiguity group effects, which specify components to be re-ranked in an ambiguity group according to an associated ranking effect. Additonally, a symptom-fault model and a failure model can be analyzed to determine symptom-fault relationships and failure modes which are applicable to the system operation. Each applicable symptom-fault relationship and failure mode is also associated with a set of ambiguity group effects which rerank the ambiguity group. A structural model is analyzed starting with the components in the ambiguity group having the greatest probability of failure. As a result of the analysis, maintenance options specifying tests to be performed on the system are output.

Claims

1. A method for diagnosing faults in a system under test comprising the steps of:

analyzing a first representation of said system under test to obtain a first list of suspect components and a ranking for each of said suspect component that indicates a level of suspicion;

pointing to a component in a structural model of said system under test according to said ranking list of suspect components;

analyzing said structural model starting at said component; and

outputting a maintenance option to be performed on said system as a result of said analysis.

2. The method as claimed in claim 1, wherein said first representation is an event based representation that defines the temporal performance of said system under test.

3. The method as claimed in claim 1, wherein said first representation is a heuristic rule based model of said system under test.

4. The method as claimed in claim 1, further comprising the steps of:

before pointing to said structural model, analyzing a second representation of said system under test to obtain a second list of suspect components and a ranking for each of said suspect components on said second list which indicates a level of suspicion;

intergrating said first list of suspect components and rankings with said second list of suspect components and rankings to obtain a integrated group of ranked suspect components.

5. The method as claimed in claim 4, wherein said first representation is an event based representation of said system under test and said second representation is a heuristic rule based model of said system under test.

6. The method as claimed in claim 1, further comprising the steps of:

grouping related components from said first list of suspect components prior to pointing to said component in said structural model.

7. The method as claimed in claim 6, wherein functionally related components are grouped together.

8. The method as claimed in claim 6, wherein structurally related components are grouped together.

9. The method as claimed in claim 1, further comprising the step of:

analyzing the result of performing said maintenance option and adjusting said list of suspect components accordingly to obtain a reranked list of suspect components;

pointing to a second component in said structural model according to said reranked list of suspect components; and

analyzing said structural model starting at said second component and outputting a new maintenance option.

10. A method for diagnosing faults in a system under test, comprising the steps of:

analyzing a first representation of said system under test to obtain a first list of suspect components and a ranking for each suspected component that indicates a level of suspicion;

analyzing a second representation of said system under test to obtain a second list of suspected components and a ranking for each suspected components that indicates a level of suspicion;

obtaining an integrated list of suspect components and a ranking for each of said suspected components from said first list and second list;

grouping related components within said integrated list of suspect components;

pointing to a component in a structural model of said system under test according to one of the grouped lists of suspect components;

analyzing said structrual model starting at said component; and

outputting a maintenance option to be performed on said system under test as a result of said analysis.

11. The method as claimed in claim 10, further comprising the step of:

analyzing the result of performing said maintenance option and adjusting said list of suspect components accordingly to obtain a reranked list of suspect components;

pointing to a second component in said structural model according to said reranked lists of components; and

analyzing said structural model starting at said second component and outputting a new maintanance option to be performed on said system.

12. A method for diagnosing faults in a system under test, comprising the steps of:

comparing a plurality of data samples collected from said system under test during its operation to data in an event based representation to recognize events which occurred in said system during its operation;

ranking a list of components in said system according to the recognized events;

pointing to component in a structural model of said system according to said ranked list of components;

analyzing said structural model starting at said component; and

outputting a maintenance option to be performed on said system as a result of analyzing said component model.

13. The method as claimed in claim 12, wherein said event based representation defines the temporal performance of said system under test.

14. The method as claimed in claim 13, wherein said event based representation of said system includes a plurality of defined events, each of which includes at least one critical parameter and wherein one of said defined events is recognized when said collected data samples match a critical parameter from said one of said plurality of defined events.

15. The method as claimed in claim 12, further comprising the steps of:

analyzing the result of performing said maintenance option and adjusting said ranking of said list of components;

pointing to a new entry point in said event structured component model according to said reranked list of components; and

analyzing said event structured component model at said new entry point and outputting a new maintenance option to be performed on said system.

16. The method as claimed in claim 12, further comprising the steps of:

grouping related components which are pointed to in said structural model prior to outputting said maintenance options.

17. The method as claimed in claim 16, wherein functionally related components are grouped together.

18. The method as claimed in claim 16, wherein structurally related components are grouped together.

19. The method as claimed in claim 12, further comprising the steps of:

prior to pointing to a component in said structural component model, comparing data observed from said system during its operation to a sympton-fault model of said system to find a subset of applicable sympton-fault relationships from said model; and

adjusting said ranking of said list of components accordingly.

20. The method as claimed in claim 19, further comprising the steps of:

prior to pointing to a component in said structural model, from said recognized events and from said collected data samples to a failure model; and

adjusting said ranking of said list of components accordingly.

21. The method as claimed in claim 12, further comprising the steps of:

prior to pointing to a component in said structural model, from said recognized events and from said collected data samples to a failure model; and

adjusting said ranking of said list of components accordingly.

22. A fault diagnostic tool for a system under test, comprising:

data acquisition means for collecting data from said system under test during its operation to obtain operational data;

an event record data base for providing data representative of a plurality of predefined events that occur during operation of said system under test;

first comparison means for comparing said operational data to said event record data base to recognize any of said predefined events that occurred during operation of said system under test;

memory means for storing a listing of a plurality of components from said system under test in order according to their probability of failure, wherein said components and said order are specified by said predefined events recognized by said comparison means;

a structural data base for providing data representative of said system under test'"'"'s structure; and

according to said listing and for outputting suggested operations to be performed on said system under test.

23. A fault diagnostic tool as claimed in claim 22, further comprising:

a first heuristic data base for providing data representative of a plurality of symptom-fault relationships for said system under test; and

second comparison means for comparing data observed from said system under test to said first heuristic data base to recognize a subset of said plurality of symptom-fault relationship and to control said listing in said memory means according to components and ranking effects associated with said recognized subset of symptom-fault relationships.

24. A fault diagnostic tool as claimed in claim 23, further comprising;

a second heuristic data base for providing data representative of a plurality of failure modes for said symptom under test; and

third comparison means for comparing said second heuristic data base to said operational data, to said predefined events recognized by said first comparison means and to said subset of sympton-fault relationships recognized by said second comparison means and to control said listing in said memory means according to components and ranks associated with any recognized failure modes.