Method for automating the development and execution of diagnostic reasoning software in products and processes

• US 5,544,308 A
• Filed: 08/02/1994
• Issued: 08/06/1996
• Est. Priority Date: 08/02/1994
• Status: Expired due to Term

• Current Assignee: VSE Corporation
• Original Assignee: Giordano Automation Corp.
• Inventors: Giordano, John P., Granieri, Michael N., Levy, Ford, deMare, Gregory, Longendorfer, Betsy, Giordano, Gerard J., Nolan, Mary E.
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Fisch, Alan M.
• Application Number:
  US08/284,382
• Time in Patent Office: 735 Days
• Field of Search: 395/575, 395/64, 395/183.02, 395/911, 395/912, 395/50, 395/51, 371/15.1, 371/16.1, 364/274.5
• US Class Current: 714/26
• CPC Class Codes: G06F 11/2257 using expert systems

Abstract

A method for automated diagnosis of faults in a system containing repairable parts is performed by selecting a set of faults representing all known failures which can occur among the parts of the system, characterized by symptom data representing the expected passing or failing results for tests applied at selected test locations in the system, generating a fault/symptom matrix of the set of faults mapped to the expected passing and failing results for the selected test locations, then performing actual tests one or more test locations and correlating the actual passing or failing test results to the fault/symptom matrix in order to identify a suspect list of faults. Additional tests may be performed until the suspect list cannot be reduced further. For efficiency, the tests are selected according to which have most diagnostic significance. The design data for the parts of the system are captured and the fault/symptom matrix is preprocessed for diagnostic efficiency and speed during run time. The diagnostic method can be implemented on a single microchip and embedded in a system for on-line diagnosis during system operation, for equipment testing, or even operator training.

Claims

1. A method for automated diagnosis of faults in a system containing reparable parts comprising the steps of:

selecting a set of faults representing all known failures which can occur among a given plurality of reparable parts contained in the system, wherein each fault is characterized by symptom data comprising expected passing or failing results for tests that can be applied to a plurality of test locations in the system;

selecting a plurality of test locations in the system at which given tests can be applied and respective outputs can be acquired as actual passing or failing results for those tests;

mapping the set of faults to their characteristic symptom data of expected passing or failing test results by generating a fault/symptom matrix defined by the set of faults as matrix row indices, by the plurality of test locations as matrix column indices, and by the expected passing and failing results for the plurality of test locations as matrix elements representing the symptom data for each of the respective faults;

performing tests at at least some of the plurality of test locations of the system, then acquiring the outputs of the performed tests as actual passing or failing test results;

correlating the actual passing or failing test results to the expected test results comprising the symptom data for the faults defined in the fault/symptom matrix; and

creating a suspect list of faults having symptom data corresponding to the actual passing or failing test results of the performed tests;

wherein said method is implemented by programmed means of a computer processing unit, said fault/symptom matrix is stored as a knowledge base of the computer processing unit, and the outputs of the tests performed on the System being diagnosed are provided as inputs to the computer processing unit;

wherein said method implemented by the programmed means of the computer processing unit is optimized for run-time efficiency by preprocessing the steps of selecting the set of faults and plurality of test locations and generating the fault/symptom matrix and storing it as a preprocessed, deterministic part of the knowledge base beforehand, then dynamically processing the steps of acquiring the outputs of the performed tests, correlating the actual test results to the stored fault/symptom matrix, and creating the suspect list of faults at run time.

2. A method for automated diagnosis of a system according to claim 1, further comprising the steps of reducing the suspect list of faults by performing additional tests at other test locations and correlating the actual test results to the symptom data for faults on the suspect list and updating the suspect list of faults reductively.

3. A method for automated diagnosis of a system according to claim 2, further comprising the steps of producing a list of repairable parts containing the faults on the suspect list, and continuing the step of reducing the suspect list of faults until the list of repairable parts contains a single repairable part or until no further test can be performed that will reduce the suspect list of faults.

4. A method for automated diagnosis of a system according to claim 1, wherein said step of creating the suspect list of faults includes, if no faults have been placed on the suspect list by correlation of actual test results to the symptom data of the fault/symptom matrix, the substep of examining each fault in turn and adding it to the suspect list if its symptom data matches at least one actual failing test result.

5. A method for automated diagnosis of a system according to claim 2, wherein the step of performing additional tests includes the preliminary substeps of scanning the list of identified tests for the system, eliminating the tests already performed, scanning each column of the fault/symptom matrix representing the other tests for those faults currently remaining on the suspect list and eliminating those tests where the columnar passing or failing symptom data for the currently remaining faults are all the same, and creating an updated list of tests to be performed on the system.

6. A method for automated diagnosis of a system according to claim 5, wherein the step of performing additional tests includes the further preliminary substep of selecting only those tests from the updated list of tests to be performed which have most diagnostic significance by scanning the fault/symptom matrix for those faults currently remaining on the suspect list, counting for each remaining fault the number of passing and failing symptom data defined for the remaining tests, and selecting only those tests where the number of passing symptom data is closest to the number of failing symptom data as being most diagnostically significant.

7. A method for automated diagnosis of a system according to claim 1, wherein said method implemented by the programmed means of the computer processing unit is optimized for run-time efficiency by preprocessing the fault/symptom matrix with a set cover analysis in which each fault row has identified with it all other fault rows having the same failing symptom data, so that the steps of correlating the test results to the faults of the fault/symptom matrix and creating the suspect list of faults can be shortened at run time.

8. A method for automated diagnosis of a system according to claim 1, further including the step of incorporating said method implemented by programmed means of the computer processing unit in a single microcontroller chip set.

9. A method for automated diagnosis of a system according to claim 1, further including the step of coupling an output of the computer processing unit to a display unit for display of information obtained by diagnosis of the system.

10. A method for automated diagnosis of a system according to claim 1, further including the step of coupling an output of the computer processing unit for transmission to a remote station.

11. A method for automated diagnosis of a system according to claim 1, further including the step of coupling an output of the computer processing unit for guidance of test probes for tests to be performed on the system.

12. A method for automated diagnosis of a system according to claim 1, wherein said steps of selecting the set of faults and plurality of test locations and generating the fault/symptom matrix includes the preliminary step of capturing design data representing the interdependencies of all of the parts of the system and converting the design data into a mapping of each possible fault to each possible test result.

13. A method for automated diagnosis of a system according to claim 1 as used for isolation identification of a repairable part of a system.

14. A method for automated diagnosis of a system according to claim 1 as used for on-line diagnosis of a system during system operation to permit automatic system reconfiguration before total system failure.

15. A method for automated diagnosis of a system according to claim 1 as used in an embedded module for built-in diagnosis of a system.

16. A method for automated diagnosis of a system according to claim 1 as used in an embedded module of test equipment for testing a system.

17. A method for automated diagnosis of a system according to claim 1 as used in an embedded module in a space system for remote transmission of diagnostic information to a ground station.

18. A method for automated diagnosis of a system according to claim 1 as used in a training mode wherein the fault/symptom matrix is used to identify expected test results given specified fault assumptions.