Method and system for diagnosing the behavior of a machine controlled by a discrete event control system

• US 5,949,676 A
• Filed: 07/30/1997
• Issued: 09/07/1999
• Est. Priority Date: 07/30/1997
• Status: Expired due to Term

• Current Assignee: Allen-Bradley Company (Rockwell Automation, Inc.)
• Original Assignee: Allen-Bradley Company LLC (Rockwell Automation, Inc.)
• Inventors: Elsley, Richard K.
• Primary Examiner(s): Grant, William
• Assistant Examiner(s): THOMAS, CAROLYN D
• Application Number:
  US08/902,737
• Time in Patent Office: 769 Days
• Field of Search: 364/184, 364/185, 364/578, 702/181, 702/185, 371/25.1, 395/183.01
• US Class Current: 700/79
• CPC Class Codes: G05B 23/0229 knowledge based, e.g. expert systems; genetic algorithms
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks

Abstract

A method and system for diagnosing the behavior of a machine which performs a discrete event process. A diagnostic engine observes occurrences of discrete events to detect timing patterns. A statistical analysis is performed on repetitions of the timing patterns, thus providing the basis for defining diagnostic rules. The rules are updated in real time and are subject to a rule competition procedure in which only the most precise and efficient rules are selected. Further occurrences of discrete events are evaluated relative to the diagnostic rules to identify and predict malfunctions of the machine.

Claims

1. A system for developing diagnostic rules for diagnosing the behavior of a machine performing a process with discrete events, the system comprising:

a main processor; and

a plurality of control elements in communication with the main processor and the machine, the control elements detecting the discrete events and communicating them to the main processor,

wherein the main processor is configured (1) to detect a timing pattern in the occurrences of the discrete events, the timing pattern including a trigger event, a result event, and a time interval between the trigger and result events, (2) to define a diagnostic rule based on a statistical analysis of repetitions of the timing pattern, and (3) to update the diagnostic rule in real time based on a detected change in the timing pattern.

2. The system as recited in claim 1, wherein the statistical analysis includes calculating a mean time interval between the trigger and result events and a standard deviation from the mean time interval.

3. The system as recited in claim 2, wherein the main processor calculates a ratio of the standard deviation to the mean time interval and defines the diagnostic rule if the magnitude of the ratio is substantially equal to or less than a predetermined maximum magnitude.

4. The system as recited in claim 2, wherein the main processor defines the diagnostic rule if the duration of the mean time interval is substantially equal to or less than a predetermined maximum duration.

5. The system as recited in claim 2, wherein the main processor defines the diagnostic rule if the magnitude of the standard deviation is substantially equal to or less than a predetermined maximum deviation.

6. The system as recited in claim 1, wherein the detected change in the timing pattern is a duration change in the time interval between the trigger and result events.

7. The system as recited in claim 1, wherein the detected change in the timing pattern is the occurrence of the result event a second time interval after a second trigger event, and the main processor (1) defines a second diagnostic rule based on a statistical analysis of repetitions of a second timing pattern including the second trigger event, the result event and the second time interval, and (2) updates the diagnostic rule by replacing the diagnostic rule with the second diagnostic rule.

8. The system as recited in claim 7, wherein the statistical analysis of the repetitions of the timing pattern includes calculating a mean time interval between the trigger event and the result event and a standard deviation from the mean time interval, and the statistical analysis of the repetitions of the second timing pattern includes calculating a second mean time interval between the second trigger event and the result event and a second standard deviation from the second mean time interval.

9. The system as recited in claim 8, wherein the main processor updates the diagnostic rule by replacing the diagnostic rule with the second diagnostic rule if the second standard deviation is smaller than the standard deviation.

10. The system as recited in claim 8, wherein the main processor determines a first ratio of the standard deviation to the mean time interval and a second ratio of the second standard deviation to the second mean time interval, and the main processor updates the diagnostic rule by replacing the diagnostic rule with the second diagnostic rule if the second ratio is smaller than the first ratio.

11. The system as recited in claim 1, wherein the machine has a first and second machine state during which discrete events occur, and wherein the timing pattern is a common timing pattern occurring during at least both the first and second machine states, and the diagnostic rule is based on a statistical analysis of repetitions of the common timing pattern.

12. The system as recited in claim 1, wherein the diagnostic rules are symmetric diagnostic rules.

13. The system as recited in claim 1, further comprising a communication link and a plurality of control modules coupled to the communication link to communicate the occurrences of the discrete events to the main processor, wherein each control module is configured to detect the transitions between the states of at least one of the control elements.

14. The system as recited in claim 1, wherein the main processor is further configured to evaluate the detected change in the timing pattern relative to the diagnostic rule to identify a malfunction in the behavior of the machine.

15. The system as recited in claim 1 wherein the main processor is configured further (4) to evaluate in real time the occurrences of the discrete events relative to the diagnostic rule to identify a malfunction in the behavior of the machine.

16. The system as recited in claim 15, wherein the real time statistical analysis includes calculating a mean time interval between the trigger and result events and a standard deviation from the mean time interval.

17. The system as recited in claim 16, wherein the main processor identifies the malfunction when the result event does not occur after the trigger event within the mean time interval plus or minus at least one standard deviation.

18. The system as recited in claim 16, wherein the real time statistical analysis further includes calculating an updated mean time interval for each timing pattern based on further repetitions of the timing pattern.

19. The system as recited in claim 18, wherein the main processor identifies the malfunction when the result event does not occur after the trigger event within the updated mean time interval plus or minus at least one standard deviation.

20. The system as recited in claim 15, wherein the main processor identifies the malfunction when an unexpected event occurs.

21. The system as recited in claim 15, wherein the system further comprises a display in communication with the main processor to provide visible indicia representative of the malfunction.

22. The system as recited in claim 15, wherein the main processor is further configured to update the diagnostic rule in real time based on a detected change in the timing pattern.

23. A method for developing diagnostic rules for diagnosing the behavior of a machine having a plurality of control elements which cooperate to perform a plurality of discrete events the method comprising the steps of:

monitoring occurrences of discrete events associated with the plurality of control elements;

detecting a timing pattern in the occurrences of the discrete events,

the timing pattern including a trigger event, a result event and a time interval between the trigger and result events;

performing a statistical analysis of repetitions of the timing pattern;

defining a diagnostic rule based on the statistical analysis;

detecting in real time a change in the timing pattern; and

updating in real time the diagnostic rule based on the detected change.

24. The method as recited in claim 23, wherein performing the statistical analysis includes calculating a mean time interval and a standard deviation from the mean time interval.

25. The method as recited in claim 24, further comprising:

calculating a ratio of the standard deviation to the mean time interval;

comparing the magnitude of the ratio to a predetermined maximum magnitude; and

rejecting the diagnostic rule if the magnitude of the ratio exceeds the predetermined maximum magnitude.

26. The method as recited in claim 24, further comprising:

comparing the duration of the mean time interval to a predetermined maximum duration; and

rejecting the diagnostic rule if the duration of the mean time interval exceeds the predetermined maximum duration.

27. The method as recited in claim 24, wherein the detected change in the timing pattern is a duration change in the time interval, and updating the diagnostic rule includes calculating an updated mean time interval based on the duration change.

28. The method as recited in claim 23, wherein the detected change in the timing pattern is the occurrence of the result event a second time interval after a second trigger event, thereby defining a second timing pattern, and updating the diagnostic rule includes:

performing a second statistical analysis of repetitions of the second timing pattern;

defining a second diagnostic rule based on the second statistical analysis; and

replacing the diagnostic rule with the second diagnostic rule.

29. The method as recited in claim 28, wherein

performing the statistical analysis includes calculating a mean time interval and a standard deviation from the mean time interval,

performing the second statistical analysis includes calculating a second mean time interval and a second standard deviation from the second mean time interval, and

replacing the diagnostic rule includes comparing the standard deviation to the second standard deviation and replacing the diagnostic rule with the second diagnostic rule if the second standard deviation is smaller than the standard deviation.

30. The method as recited in claim 23, wherein the machine has a first machine state for performing a first discrete event process and a second machine state for performing a second discrete event process, the method further comprising:

selecting a common timing pattern occurring during at least both the first and second machine states;

performing a statistical analysis of repetitions of the common timing pattern; and

defining a common diagnostic rule based on the statistical analysis.

31. The method as recited in claim 23, further comprising evaluating the detected change in the timing pattern relative to the diagnostic rule and identifying a malfunction in the behavior of the machine based on the evaluation.

32. The method as recited in claim 23 further including the step of

identifying in real time a malfunction in the behavior of the machine based on evaluating the occurrences of the discrete events relative to the diagnostic rule.

33. The method as recited in claim 32, wherein performing the real time statistical analysis includes calculating a mean time interval and a standard deviation from the mean time interval.

34. The method as recited in claim 33, wherein identifying the malfunction includes evaluating whether the result event occurs after the trigger event within the mean time interval plus or minus at least one standard deviation.

35. The method as recited in claim 33, wherein performing the real time statistical analysis further includes calculating an updated mean time interval for each timing pattern based on further repetitions of the timing pattern.

36. The method as recited in claim 32, wherein identifying the malfunction includes evaluating whether an unexpected event has occurred.

37. The method as recited in claim 32, further comprising displaying visible indicia representative of the malfunction.

38. The method as recited in claim 32, further comprising updating the diagnostic rule in real time based on a detected change in the timing pattern.