Methods and apparatus for early fault detection and alert generation in a process

• US 20030225466A1
• Filed: 05/30/2002
• Published: 12/04/2003
• Est. Priority Date: 05/30/2002
• Status: Active Grant

• Current Assignee: ADA ANALYTICS ISRAEL LTD (Advanced Emissions Solutions, Inc.)
• Original Assignee: Insyst, Ltd. (Advanced Emissions Solutions, Inc.)
• Inventors: Fisher, Yossi, Seror, Jacques, Hartman, Jehuda, Yulevitch, Oren, Peretz, Joseph
• Granted Patent: US 7,096,074 B2
• Time in Patent Office: Days
• US Class Current: 700/80
• 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
• CPC Class Codes: G05B 23/0267 Fault communication, e.g. human machine interface [HMI]
• CPC Class Codes: G05B 23/0283 Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL] preventive maintenance, i.e. planning maintenance according to the available resources without monitoring the system G06Q10/06
• CPC Class Codes: G06F 18/2433 Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection

Abstract

A method and apparatus for providing early fault detection and alert generation in a multi-parameter process is provided. The method comprises defining a multi-dimensional space, each dimension representing a parameter of the process; applying an analysis of the process to the space in order to define regions of high and low probability of occurrence; measuring and mapping parameter values onto the space; and generating an alert whenever a parameter value is mapped into a region of low probability. The method and apparatus are also useful for identifying the cause of faults, adjusting the process to avoid faults and predicting a process-specific maintenance schedule.

Claims

1. A method of providing an early warning of an abnormal situation in a process, said process being subject to influence by at least two parameters, said method comprising:

a. defining a space representing said process, said space comprising at least two dimensions, each dimension representing a parameter of said process;

b. applying an analysis of parameter values of said process to said space in order to define regions of high and low probability of occurrence within said space;

c. measuring said parameters to determine current values thereof;

d. mapping said measured parameter values onto said space; and

e. generating a signal indicating an abnormal situation whenever said measured parameter values are mapped into a region of low probability of occurrence.

2. The method of claim 1, further comprising generating a signal indicating an abnormal situation whenever said mapped parameter values indicate drifting toward a region of lower probability.

3. The method of claim 1, further comprising dividing said parameter values into discrete value segments and dividing said space into discrete areas, and using said value segments to assign priority values to each of said areas.

4. The method of claim 1, wherein said application of analysis of parameter values is ongoing and at least some of said regions of high and low probability of occurrence are continually adjusted.

5. The method of claim 1, wherein at least some of said regions of high and low probability of occurrence are predesignated.

6. The method of claim 1, further comprising providing associations between said parameters by allowing a user to

a. select a mapped parameter;

b. examine selected ones of associated parameters;

c. examine further selected parameters associated with said selected parameters;

thereby to identify causes of said abnormal situations.

7. The method of claim 1, wherein said process is subject to influence by at least three parameters, and said space comprises at least three dimensions, each dimension representing a parameter of said process.

8. The method of claim 1, wherein said process is subject to influence by four or more parameters, and said space comprises four or more dimensions, each dimension representing a parameter of said process.

9. The method of claim 1, wherein said process is subject to influence by ten or more parameters, and said space comprises ten or more dimensions, each dimension representing a parameter of said process.

10. The method of claim 1, wherein said process is subject to influence by at least one virtual parameter.

11. The method of claim 1, wherein said parameters comprise at least one of a group consisting of process inputs, process outputs and intermediate values derived from said process.

12. The method of claim 1, wherein a relationship between said parameters is non-linear.

13. The method of claim 1, wherein said analysis is an empirical analysis.

14. The method of claim 1, wherein said analysis is a statistical analysis.

15. The method of claim 1, wherein said signal is visible or audible.

16. The method of claim 1, wherein said signal is an interruption of said process.

17. The method of claim 1, wherein said signal is a variation of said process.

18. A method of developing a model of a multi-parameter process for the purpose of providing an early warning of an abnormal situation in said process, said developing comprising:

a. defining a space representing said process, said space having a dimension representing each parameter of said process;

b. extracting values for each of said parameters;

c. mapping said extracted parameter values onto said space; and

d. applying an analysis of said mapped parameter values in order to define regions of high and low probability of occurrence within said space;

such that said regions of high and low probability of occurrence provide a model of said process.

19. The method of claim 18, wherein said extracting and mapping of parameter values is ongoing, said model being variable according to subsequent mapping of said parameter values.

20. The method of claim 18, wherein said parameters comprise at least one of a group consisting of process inputs, process outputs and intermediate values derived from said process.

21. The method of claim 18, wherein a relationship between said parameters is non-linear.

22. The method of claim 18, wherein said analysis is an empirical analysis.

23. The method of claim 18, wherein said analysis is a statistical analysis.

24. A method of providing an early warning of an abnormal situation in a multi-parameter process, said method using a model of said process, said model comprising a defined space representing said process, said space having defined therein regions of high and low probability of occurrence of values of said parameters said method comprising

a. measuring said parameters to determine current values thereof;

b. forming said current values into process vectors for mapping;

c. mapping said process vectors onto said model; and

d. generating a signal indicating an abnormal situation whenever said process vectors are mapped into a region of low probability.

25. The method of claim 24, wherein a range for said current values is divided into at least a high and low region.

26. The method of claim 24, wherein said parameters comprise at least one of a group consisting of process inputs, process outputs and intermediate values derived from said process.

27. The method of claim 24, further comprising generating a signal indicating an abnormal situation whenever said process vectors are mapped drifting toward a region of lower probability.

28. The method of claim 24, wherein said signal is visible or audible.

29. The method of claim 24, wherein said signal is an interruption of said process.

30. The method of claim 24, wherein said signal is a variation of said process.

31. In a process subject to influence by at least two parameters, each of said parameters having a range of discrete values, a method of providing an early warning of an abnormal situation in said process caused by at least one parameter of said process, said method comprising

a. defining a space representing said process, said space comprising at least two dimensions, each dimension representing a parameter of said process and being divided into discrete areas corresponding to said discrete values of said parameter, thereby to discretize said space;

b. applying an empirical analysis of said process to said discretized space in order to define discrete regions of high and low input probability within said space, or to define normal and abnormal values within said range of discrete values of at least one of said parameters;

c. measuring said parameters to determine current values thereof and forming process vectors based upon said measured parameter values;

d. mapping said process vectors onto said space; and

e. generating a signal indicating an abnormal situation whenever said process vectors are mapped into a discrete region of low probability.

32. The method of claim 31, further comprising generating a signal indicating an abnormal situation whenever said process vectors are mapped drifting toward a discrete region of lower probability.

33. The method of claim 31, wherein said application of said empirical analysis is ongoing and said discrete regions of high and low input probability are continually adjusted.

34. The method of claim 31, wherein the mapping of the process vectors of at least some of said parameters onto said space is used to determine a maintenance schedule for the process.

35. The method of claim 34, wherein a probability calculation is used to determine frequency of maintenance.

36. The method of claim 34, wherein said mapping is repeatedly carried out, the movement of the process vectors within said space being used to determine frequency of maintenance.

37. The method of claim 31, wherein said process is subject to influence by at least three parameters, and said space comprises at least three dimensions, each dimension representing a parameter of said process.

38. The method of claim 31, wherein said process is subject to influence by four or more parameters, and said space comprises four or more dimensions, each dimension representing a parameter of said process.

39. The method of claim 31, wherein said process is subject to influence by ten or more parameters, and said space comprises ten or more dimensions, each dimension representing a parameter of said process.

40. The method of claim 31, wherein said process is subject to influence by at least one virtual parameter.

41. The method of claim 31, wherein said parameters comprise at least one of a group consisting of process inputs, process outputs and intermediate values derived from said process.

42. The method of claim 31, wherein the relationship between said parameters is non-linear.

43. The method of claim 31, wherein said signal is visible or audible.

44. The method of claim 31, wherein said signal is an interruption of said process.

45. Apparatus for generating an early warning of an abnormal situation in a process, said process being subject to influence by at least two parameters, said apparatus comprising

a. a learning unit for developing a model of said process, said learning unit comprising

(i) an analysis unit for defining a space representing said process, said space having at least two dimensions, each dimension representing a parameter of said process;

(ii) at least one extracting unit for extracting values relating to said parameters; and

(iii) a mapping unit for mapping said parameter values onto said space;

 

said analysis unit being further for analyzing said mapping in order to divide said space into regions of high and low probability of occurrence, thereby to form said model; and

b. an application unit comprising

(i) at least one parameter measuring unit for measuring at least one parameter to determine at least one parameter value and to form said at least one parameter value into a process vector; and

(ii) a mapping unit for mapping said process vector onto said model; and

(iii) an alert unit for generating a signal when a process vector is mapped onto a region of low probability.

46. The apparatus of claim 45, wherein said analyzing said mapping by said analysis unit is ongoing in order to continually adjust said regions of high and low probability of occurrence.

47. The apparatus of claim 45, wherein said process is subject to influence by at least three parameters, and said space comprises at least three dimensions, each dimension representing a parameter of said process.

48. The apparatus of claim 45, wherein said process is subject to influence by four or more parameters, and said space comprises four or more dimensions, each dimension representing a parameter of said process.

49. The apparatus of claim 45, wherein said process is subject to influence by ten or more parameters, and said space comprises ten or more dimensions, each dimension representing a parameter of said process.

50. The apparatus of claim 45, wherein said process is subject to influence by at least one virtual parameter.

51. The apparatus of claim 45, wherein said parameters comprise at least one of a group consisting of process inputs, process outputs and intermediate values derived from said process.

52. The apparatus of claim 45, wherein said extracting unit is further for dividing said parameters into discrete value segments and said analysis unit is further for dividing said space into discrete areas and using said value segments to assign priority values to each of said areas.

53. The apparatus of claim 45, further comprising a maintenance scheduling unit within said application unit for monitoring said parameter values in order to schedule maintenance of said process.

54. The apparatus of claim 45, further comprising a diagnostic unit within said application unit for providing associations between said parameters, said diagnostic unit allowing a user to

a. select a mapped parameter;

b. examine selected ones of associated parameters; and

c. examine further selected parameters associated with said selected parameters;

thereby to identify causes of abnormal situations in said process.

55. The apparatus of claim 45, further comprising a process adjustment unit within said application unit for using said regions of high and low probability of occurrence to dictate adjustment of said process via at least one of said parameters, thereby to avoid said abnormal situation.