Fault finder expert system

• US 5,214,653 A
• Filed: 10/22/1990
• Issued: 05/25/1993
• Est. Priority Date: 10/22/1990
• Status: Expired due to Term

• Current Assignee: Harris Corporation (L3Harris Technologies, Inc.)
• Original Assignee: Harris Corporation (L3Harris Technologies, Inc.)
• Inventors: Schneider, Mordechay, Elliott, William M. Jr.
• Primary Examiner(s): ATKINSON, CHARLES
• Application Number:
  US07/600,831
• Time in Patent Office: 946 Days
• Field of Search: 371/15.1, 371/16.1, 364/274.5, 364/274.6, 364/274.7, 364/276.3, 395/53, 395/61, 395/77, 395/916, 395/917, 395/575
• US Class Current: 714/26
• CPC Class Codes: G01R 31/2846 using hard- or software simulation or using knowledge-based systems, e.g. expert systems, artificial intelligence or interactive algorithms
• CPC Class Codes: G06F 11/2257 using expert systems
• CPC Class Codes: Y10S 706/916 Electronic or computer, internal or network, circuit

Abstract

A device for finding faults in a target system that has a plurality of "LRUs", which are the lowest replaceable units in the target system. The fault finding device receives input data relating to a state of the target system, stores a probability of failure of individual LRUs of the target system, and stores rules, these rules relating conclusions and premises, with at least some of the conclusions identifying a faulty LRU. The device determines a certainty of a conclusion of a stored rule in response to a comparison of the premises of the rule and the input data, and orders the conclusions of the stored rules as a function of both the probability of failure of individual LRUs and the certainties of conclusions of the stored rules.

Claims

1. A device for finding faults in a target system that has a plurality of LRUs, comprising:

means for receiving input data relating to a state of the target system;

means for storing a probability of failure of individual LRUs of the target system;

means for storing rules, said rules relating conclusions and premises, at least some of said conclusions identifying a faulty LRU;

means for determining a certainty of a conclusion of a stored rule in response to the comparison of the premises of the rule and the input data;

means for ordering the conclusions of the stored rules as a function of both the probability of failure of individual LRUs and the certainties of conclusions of the stored rules; and

means for providing said ordered conclusions to a user via a computer-user interface to thereby indicate a most likely LRU to be a cause of the fault in the target system.

2. The device of claim 1, wherein the means for storing rules is a knowledge base coupled to the means for determining certainties, said knowledge base containing stored rules having premises and conclusions, each rule having associated:

a value representing a number of times the rule was tried;

a value representing a number of times the rule was successful;

an expert certainty value representing a human expert'"'"'s certainty in correctness of the rule;

a current certainty value representing a current certainty in correctness of the rule as a function of the number of times the rule was tried and the number of times the rule was successful.

3. The device of claim 2, wherein the means for storing a probability of failure is an updatable LRU data base in which a probability of failure for each of the LRUs in the target system is provided.

4. The device of claim 3, wherein said means for receiving input data includes a blackboard having an output coupled to the means for determining a certainty of a conclusion, said blackboard providing said input data to said means for determining a certainty of a conclusion.

5. The device of claim 4, wherein the input data contained in the blackboard and the rules stored in the knowledge base are parsed.

6. The device of claim 5, further comprising means for updating the current certainty value of a rule in the knowledge base when the correctness of the conclusion of that rule has been verified by a user.

7. The device of claim 6, wherein the means for updating the current certainty value updates the current certainty value according to the equation:

Rcert;

=Rcert-0.5 * (Rcert-NS/NT), where Rcert is the current certainty value, NS is the value representing the number of times the rule was successful, and NT is the value representing the number of times the rule was tried.

8. The device of claim 7, further comprising a status interface coupled to the blackboard, said status interface providing input data from the target system directly to the blackboard.

9. The device of claim 8, wherein the means for matching input data to the stored rules includes means for performing fuzzy matching of the input data to the stored rules to produce a match value of the match of input data to a rule.

10. The device of claim 9, wherein the means for matching input data to the stored rules further includes means for firing a particular rule if and only if the means for performing fuzzy matching produces a match value above a threshold value.

11. The device of claim 10, wherein the threshold value is 0.6.

12. The device of claim 10, further comprising means for updating the blackboard to contain a firing certainty for a particular rule as a function of the match value, the expert certainty value and the current certainty value for that particular rule.

13. The device of claim 12, wherein the means for ordering orders the conclusions according to a predefined classing of combinations of probabilities of failure and certainties of conclusion.

14. The device of claim 13, further comprising means for combining the probabilities of failure of LRUs with the certainties of conclusion of rules to provide final firing certainties for two or more rules when two or more conclusions are placed into the same class by the means for ordering.

15. The device of claim 14, wherein the means for combining includes means for weighting the probability of failure and the certainty of conclusion of a rule and summing the weighted probability of failure and the weighted certainty of conclusion of a rule to produce a final firing certainty of a rule.

16. The device of claim 15, further comprising an explanation module coupled to the means for ordering, the explanation module providing a user with rules that are not concluding rules and which need more input data in order to reach a conclusion.

17. The device of claim 16, wherein the means for determining certainties and the means for ordering include means for performing direct chaining of the stored rules.

18. A method of using a computer for finding faults in a target system that has a plurality of LRUs, comprising:

the computer receiving input data relating to a state of the target system;

storing in a memory of the computer a probability of failure of individual LRUs of the target system;

maintaining in the memory of the computer a set of stored rules, said rules relating conclusions and premises, at least some of said conclusions identifying a faulty LRU;

the computer comparing the premises of the stored rules and the input data;

the computer determining a certainty of a conclusion of the stored rule in response to the comparison of the premises of the rule and the input data;

the computer ordering the conclusions of the stored rules as a function of both the probability of failure of individual LRUs and the certainties of conclusions of the stored rules; and

the computer providing said ordered conclusions to a user via a computer-user interface to thereby indicate a most likely LRU to be a cause of the fault in the target system.

19. The method of claim 18, wherein the stored rules are maintained in a knowledge base said knowledge base containing stored rules having premises and conclusions, each rule having associated:

a value representing a number of times the rule was tried;

a value representing a number of times the rule was successful;

an expert certainty value representing a human expert'"'"'s certainty in correctness of the rule;

a current certainty value representing a current certainty in correctness of the rule as a function of the number of times the rule was tried and the number of times the rule was successful.

20. The method of claim 19, wherein the probability of failure for each of the LRUs in the target system is stored in an updatable LRU data base.

21. The method of claim 20, wherein the input data is received in a blackboard, said blackboard providing said input data for the determining of a certainty of a conclusion.

22. The method of claim 21, wherein the input data contained in the blackboard and the rules stored in the knowledge base are parsed.

23. The method of claim 22, further comprising the step of updating the current certainty value of a rule in the knowledge base when the correctness of the conclusion of that rule has been verified by a user.

24. The method of claim 23, wherein the current certainty value is updated according to the equation:

Rcert;

=Rcert-0.5 * (Rcert-NS/NT), where Rcert is the current certainty value, NS is the value representing the number of times the rule was successful, and NT is the value representing the number of times the rule was tried.

25. The method of claim 24, further comprising the step of providing input data from the target system directly to the blackboard via a status interface.

26. The method of claim 25, wherein the step of comparing includes input data to the premises of the stored rules includes the step of performing fuzzy matching of the input data to the premise of the stored rules to produce a match value.

27. The method of claim 26, further comprising the step of firing a particular rule if and only if the step of performing fuzzy matching produces a match value above a threshold value.

28. The method of claim 27, wherein the threshold value is 0.6.

29. The method of claim 27, further comprising the step of updating the blackboard to contain a firing certainty for a particular rule as a function of the match value, the expert certainty value and the current certainty value for that particular rule.

30. The method of claim 29, wherein the conclusions are ordered according to a predefined classing of combinations of probabilities of failure and certainties of conclusion.

31. The method of claim 30, further comprising the step of combining the probabilities of failure of LRUs with the certainties of conclusion of rules to provide final firing certainties for two or more rules when two or more conclusions are placed into the same class by the step of ordering the conclusions.

32. The method of claim 31, wherein the step of combining includes the step of weighting the probability of failure and the certainty of conclusion of a rule and summing the weighted probability of failure and the weighted certainty of conclusion of a rule to produce a final firing certainty of a rule.

33. The method of claim 32, further comprising the step of providing a user with rules that are not concluding rules and which need more input data in order to reach a conclusion.