Intergrated qualitative/quantitative reasoning with enhanced core predictions and extended test procedures for machine failure isolation using qualitative physics

• US 5,522,014 A
• Filed: 04/26/1994
• Issued: 05/28/1996
• Est. Priority Date: 04/26/1994
• Status: Expired due to Term

• Current Assignee: Sikorsky Aircraft Corporation (Martin Marietta Corporation)
• Original Assignee: United Technologies Corporation (Rtx Corporation)
• Inventors: Clark, Robert T., Hamilton, Thomas P., Gallo, Steven
• Primary Examiner(s): Downs, Robert W.
• Application Number:
  US08/233,280
• Time in Patent Office: 763 Days
• Field of Search: 395/10, 395/51, 395/911, 395/912, 395/914, 395/915, 364/551.01, 364/551.02
• US Class Current: 706/45
• CPC Class Codes: G06F 11/2257 using expert systems

Abstract

The presence of particular faults in a machine is determined using constraint suspension and Qualitative Physics by estimating a time for performance for each of a plurality of machine test procedures and providing a summation, for each test procedure, of estimated times for performing portions of each of the test procedures. Portions of the machine test procedures which are unnecessary due to a present test configuration of the machine are not included in the summation. A particular test procedure is chosen from the group of machine test procedures by selecting the one of the machine test procedures having a highest test score where the test score varies inversely to the summation. The test procedure can be a component health assessment and/or an observation of a machine parameter.

Claims

1. A method of using a processor to determine the presence of particular faults in a machine, comprising the steps of:

the processor estimating a time for performance for each of a plurality of machine test procedures by providing a summation, for each test procedure, of estimated times for performing portions of each of said test procedures, wherein portions of said machine test procedures which are unnecessary due to a present test configuration of the machine are not included in said summation;

the processor choosing a particular test procedure from said group of machine test procedures by selecting the one of said machine test procedures having a highest test score, said test score being inversely proportional to said summation;

the processor prompting a user to perform said particular test procedure and provide results thereof to the processor;

the processor generating a plurality of pending hypotheses wherein each hypothesis has associated therewith a set of confluences using variables and equations to describe operation of components of the machine which are assumed not to have failed;

the processor propagating values indicative of said test results through said confluences to produce a set of predictions for values of confluence variables;

the processor discarding hypotheses which produce an inconsistent set of predictions;

the processor saving hypotheses which produce a consistent set of predictions; and

the processor indicating the presence of one or more particular machine faults in response to there remaining a single hypothesis corresponding to failure of one or more particular machine components.

2. A method of using a processor to determine the presence of particular faults in a machine, according to claim 1, wherein said prompting step includes the step of:

prompting the user to perform a component health assessment and enter the results thereof.

3. A method of using a processor to determine the presence of particular faults in a machine, according to claim 1, wherein said prompting step includes the step of:

prompting the user to perform an observation of a machine parameter and enter the results thereof.

4. A method of using a processor to determine the presence of particular faults in a machine, comprising the steps of:

the processor prompting a user to perform one of a plurality of machine test procedures and provide results thereof to the processor, said test procedure being created by applying a generic template to at least one of;

a specific component and specific parameter;

the processor generating a plurality of pending hypotheses having associated therewith a set of confluences using variables and equations to describe operation of components of the machine which are assumed not to have failed;

the processor propagating values indicative of said test results through said confluences to produce a set of predictions for values of confluence variables;

the processor discarding hypotheses which produce an inconsistent set of predictions;

the processor saving hypotheses which produce a consistent set of predictions; and

the processor indicating the presence of one or more particular machine faults in response to there remaining a single hypothesis corresponding to failure of one or more particular machine components.

5. A method of using a processor to determine the presence of particular faults in a machine, according to claim 4, wherein said prompting step includes the step of:

prompting the user to perform a component health assessment and enter the results thereof.

6. A method of using a processor to determine the presence of particular faults in a machine, according to claim 4, wherein said prompting step includes the step of:

prompting the user to perform an observation of a machine parameter and enter the results thereof.

7. A method of using a processor to determine the presence of particular faults in a machine, comprising the steps of:

the processor receiving machine signals indicative of machine test procedure results;

the processor generating a first set of pending hypotheses having associated therewith a set of confluences using variables and equations to describe operation of components of the machine which are assumed not to have failed;

the processor propagating said machine signals through said confluences to produce a set of predictions for values of confluence variables of a particular hypothesis by the steps of;

selecting a subset of said variables, said subset containing unrestricted and partially restricted confluence variables of said hypothesis;

temporarily assigning possible qualitative values to each variable of said subset to determine if one or more of said possible qualitative values results in a consistent set of predictions for said particular hypothesis; and

restricting each variable of said subset to values which produce a consistent set of predictions;

the processor discarding hypotheses which produce an inconsistent set of predictions;

the processor saving hypotheses which produce a consistent set of predictions; and

the processor indicating the presence of one or more particular machine faults in response to there remaining a single hypothesis corresponding to failure of one or more particular machine components.

8. A method of using a processor to determine the presence of particular faults in a machine, according to claim 7, wherein said selecting step includes the step of:

selecting variables that are likely to at least one of;

be restricted and cause other variables to be restricted.

9. A method of using a processor to determine the presence of particular faults in a machine, according to claim 8, wherein said selecting step includes the step of:

the processor examining said unrestricted and partially restricted variables of said particular hypothesis and selecting variables that satisfy at least one of the following conditions;

the variable appears in a precondition of a confluence, the variable appears in an unconditional confluence containing greater than two variables that are one of unrestricted and partially restricted, and the variable appears in a confluence postcondition with more than two variables that are one of unrestricted and partially restricted.

10. A method of using a processor to determine the presence of particular faults in a machine, according to claim 9, further comprising the step of:

the processor generating additional pending hypotheses having associated therewith a set of confluences using variables and equations to describe operation of components of the machine which are assumed not to have failed, said additional hypotheses corresponding to substructure of at least one of said first set of hypotheses, wherein said examining step is only performed on variables of said additional hypotheses than do not appear in said at least one of said first set of hypotheses corresponding to said additional hypotheses.

11. A method of using a processor to determine the presence of particular faults in a machine, according to claim 10, further comprising the step of:

the processor assigning values to variables of said additional hypotheses, said values being derived from values of corresponding variables from said at least one of said first set of hypotheses corresponding to said additional hypotheses.

12. A method of using a processor to determine the presence of particular faults in a machine, according to claim 10, wherein said selecting step includes the step of:

the processor examining additional variables of said additional hypotheses, said additional variables corresponding to variables from said at least one of said first set of hypotheses corresponding to said additional hypotheses, wherein said additional variables are one of;

variables that have been restricted and recursive covariates thereof.

13. A method of using a processor to determine the presence of particular faults in a machine, according to claim 10, wherein said selecting step includes the step of:

the processor examining additional variables of said additional hypotheses, said additional variables corresponding to recursive covariates of variables of said additional hypotheses that do not appear in said at least one of said first set of hypotheses corresponding to said additional hypotheses.

14. A method of using a processor to determine the presence of particular faults in a machine, according to claim 13, further comprising the steps of:

the processor receiving additional machine signals indicative of machine test procedure results; and

the processor performing said examining step only on variables and covariates thereof corresponding to said additional machine signals and implicants thereof.

15. A method of using a processor to determine the presence of particular faults in a machine, according to claim 7, wherein said propagating step further includes the steps of:

the processor determining implicants of each value assigned to a variable of said subset of variables; and

the processor restricting other variables of said confluences according to a union of said implicants.

16. A method of using a processor to determine the presence of particular faults in a machine, according to claim 7, wherein said selecting step includes the step of:

the processor examining only said unrestricted and partially restricted confluence variables and covariates thereof that had been restricted on a most recent previous iteration.

17. A method of using a processor to determine the presence of particular faults in a machine, according to claim 7, further including the step of:

the processor discarding said particular hypothesis in response to there being no possible values that can be assigned to a variable of said subset of variables resulting in a consistent set of predictions.

18. A method of using a processor to determine the presence of particular faults in a machine, comprising the steps of:

the processor receiving machine signals indicative of machine test procedure results;

the processor generating a first set of pending hypotheses having associated therewith a set of confluences using variables and equations to describe operation of components of the machine which are assumed not to have failed;

the processor assigning to confluence variables of a particular hypothesis values that correspond to said machine signals;

the processor propagating values for fully restricted and partially restricted confluence variables through said confluences to further restrict values of other variables of said confluences;

the processor discarding hypotheses which produce an inconsistent set of predictions;

the processor saving hypotheses which produce a consistent set of predictions; and

the processor indicating the presence of one or more particular machine faults in response to there remaining a single hypothesis corresponding to failure of one or more particular machine components.

19. A method of using a processor to determine the presence of particular faults in a machine, comprising the steps of:

the processor receiving machine signals indicative of machine test procedure results;

the processor generating a first set of pending hypotheses having associated therewith a set of confluences using variables and equations to describe operation of components of the machine which are assumed not to have failed;

the processor propagating said machine signals through said confluences to produce a set of predictions for values of confluence variables of a particular hypothesis, wherein at least one of said values is a quantitative value;

the processor discarding hypotheses which produce an inconsistent set of predictions;

the processor saving hypotheses which produce a consistent set of predictions;

the processor indicating the presence of one or more particular machine faults in response to there remaining a single hypothesis corresponding to failure of one or more particular machine components; and

the processor accessing a global landmark ordering to correlate quantitative values and qualitative values of said confluence variables.