Provision of process related information

• US 7,213,174 B2
• Filed: 06/05/2002
• Issued: 05/01/2007
• Est. Priority Date: 06/05/2001
• Status: Expired due to Term

• Current Assignee: ABB Schweiz AG (ABB Limited)
• Original Assignee: ABB AB (ABB Limited)
• Inventors: Dahlquist, Erik, Weidl, Galia
• Primary Examiner(s): Baderman; Scott
• Assistant Examiner(s): ASSESSOR, BRIAN J
• Application Number:
  US10/479,664
• Publication Number:
  US 20040250166A1
• Time in Patent Office: 1,791 Days
• Field of Search: 714/37, 714/38
• US Class Current: 714/37
• CPC Class Codes: G05B 23/0278 Qualitative, e.g. if-then rules; Fuzzy logic; Lookup tables; Symptomatic search; FMEA

Abstract

A method and an analyzer for producing information associated with a process is disclosed. Information from information sources that associate with the process is processed to classify the information into characteristic states. Results of the classification are then analyzed for producing information associated with at least one root cause for an event by means of at least one analyzing technology that is capable of reasoning under at least one uncertainty.

Claims

1. A method of producing information associated with a process, the method comprising:

receiving information, including time series of sampled signal points, from sources in association with the process;

determining a trend based on said received information;

classifying said information into characteristic states based on the determined trend, the classifying including classifying alarm signals between real states and false states;

mapping said time series of sampled signal points into time series of signal states based on said classification; and

analyzing said time series of signal states for producing information associated with a root cause for an event by means of at least one analyzing technology that is capable of reasoning under at least one uncertainty.

2. The method according to claim 1, wherein the event comprises an abnormal condition.

3. The method according to claim 2, wherein the event comprises a failure.

4. The method according to claim 1, wherein the classification reduces the number of variable states that are input into the analyzing stage.

5. The method according to claim 1, wherein information comprising numeric variables is classified into characteristic states based on the value of the variables.

6. The method according to claim 5, wherein a value is classified into a characteristic state that defines a range of values said value belongs to.

7. The method according to claim 1, wherein at least one characteristic state is expressed by means of a binary value.

8. The method according to claim 7, wherein information input into the analyzing stage comprises a string of binary values.

9. The method according to claim 1, wherein information input into the analyzing stage comprises a continuous distribution of variables.

10. The method according to claim 1, wherein at least one characteristic state indicates a condition.

11. The method according to claim 1, wherein alarm signals are classified into root cause alarms and effect alarms, and only root cause alarms are presented to a user.

12. The method according to claim 1, wherein said processing further comprises predicting development and/or trend of an information signal.

13. The method according to claim 1, wherein information from information sources is processed adaptively.

14. The method according to claim 1, further comprising:

use of structured information associated with the process.

15. The method according to claim 14, wherein at least a part of the structured information is acquired by means of lower level diagnostic packages.

16. The method according to claim 14, wherein said structured information comprises at least one model describing the process.

17. The method according to claim 16, wherein a model describes physics of the process.

18. The method according to claim 16, wherein a model describes mathematical characteristics of the process.

19. The method according to claim 16, wherein a model is based on statistical information or experience from the process.

20. The method according to claim 1, further comprising:

training of the information processing stage based on information about at least one trend or development determined for the process.

21. The method according to claim 1, further comprising:

use of at least one neural network for said processing of information prior the results of said processing are input into the analyzing stage.

22. The method according to claim 21, wherein at least one neuro-fuzzy network is used.

23. The method according to claim 1, further comprising:

use of principal component analysis and/or artificial intelligence agents.

24. The method according to claim 1, wherein said capability to reason under at least one uncertainty is provided by adaptive root cause analysis.

25. The method according to claim 24, wherein a hybrid combination of at least two different adaptive root cause analysis methods is used for the reasoning under at least one uncertainty.

26. The method according to claim 24, wherein the adaptive root cause analysis is provided by means of at least one of the following methods:

Bayesian inference;

neuro-fuzzy inference;

case based reasoning.

27. The method according to claim 26, wherein the cased based reasoning is for identifying an abnormality, the Bayesian inference is for determination of at least one root cause behind the abnormality, and the neuro-fuzzy inference is for analyzing abnormalities between stages and/or phase transitions in the process.

28. The method according to claim 1, further comprising:

comparing of simulated or computed data with data obtained from the process.

29. The method according to claim 28, further comprising:

trending a deviation determined by said comparing.

30. The method according to claim 1, further comprising:

normalization of information on which the analysis is to be based on.

31. The method according to claim 1, further comprising:

testing for fault and/or abnormality patterns after said classification.

32. The method according to claim 1, further comprising:

monitoring for development of a root cause for an event.

33. The method according to claim 1, wherein at least a part of the information for the analysis is provided by a distributed control system.

34. The method according to claim 1, wherein the output of the analyzing stage is used in real-time control of the process.

35. The method according to claim 1, further comprising:

prediction of required actions based on the output of the analyzing stage.

36. The method according to claim 34, wherein control actions are ranked based on at least one of the following list:

probabilities;

membership grade;

minimal error measure of similarity.

37. The method according to claim 1, wherein information associated with a plurality of root causes is produced, and said root causes are ranked based on the probabilities thereof.

38. The method according to claim 1, wherein a part of the analysis is carried out by a data processing facility in a location which is remote with respect to the physical location of at least one part of the process.

39. The method according to claim 1, wherein an operator may input feedback after presentation of the results of the analysis, and updated results are provided based on the feedback.

40. The method according to claim 1, wherein the characteristic states comprise decreasing, constant, or increasing.

41. An analyzer for analyzing a process, the analyzer comprising:

processing means adapted to receive information, including time series of sampled signal points, from sources in association with the process, to determine a trend based on said received information, to classify said information into characteristic states based on the determined trend, including classifying alarm signals between real states and false states, and to map said time series of sampled signal points into time series of signal states based on said classification; and

analyzing means for classifying information associated with the process into characteristic states and analyzer means for producing information associated with a root cause of an event by means of at least one analyzing technology that is capable of reasoning under at least one uncertainty.

42. The method according to claim 41, further comprising:

prediction means for predicting development and/or trend of an information signal.

43. The method according to claim 41, wherein the processing means and/or the prediction means are implemented by means of at least one neural network.

44. The method according to claim 43, further comprising:

at least one neuro-fuzzy network.

45. The method according to claim 41, further comprising:

a storage means for storing structured information associated with the process.

46. The method according to claim 45, wherein the structured information comprises at least one model describing the process and/or the physics of the process.

47. The method according to claim 41, wherein the analyzer means comprises means for implementing adaptive root cause analysis.

48. The method according to claim 47, wherein the analyzer means comprises means for hybrid implementation of at least two different root cause analysis methods.

49. The method according to claim 41, further comprising:

a graphical user interface.

50. The analyzer according to claim 41, further comprising:

means for communicating a message to a remote operator, said means being based on a communication media that is based on use of Internet protocol and/or a wireless communication protocol.

51. The analyzer according to claim 41, further comprising:

at least one database for storing information associated with components, spare parts and similar, and an association between the analyzer and said at least one database.

52. The analyzer according to claim 41, wherein the characteristic states comprise decreasing, constant, or increasing.

53. A computer program product, comprising:

a computer readable medium; and

program code means recorded on the computer readable medium and executable by a processor for performing any of steps of

receiving information, including time series of sampled signal points, from sources in association with the process;

determining a trend based on said received information;

classifying said information into characteristic states based on the determined trend, the classifying including classifying alarm signals between real states and false states;

mapping said time series of sampled signal points into time series of signal states based on said classification; and

analyzing said time series of signal states for producing information associated with a root cause for an event by means of at least one analyzing technology that is capable of reasoning under at least one uncertainty.