ON-LINE MULTIVARIATE ANALYSIS IN A DISTRIBUTED PROCESS CONTROL SYSTEM

• US 20080082194A1
• Filed: 03/20/2007
• Published: 04/03/2008
• Est. Priority Date: 09/29/2006
• Status: Active Grant

• Current Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Original Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Inventors: Miller, John P., SAMARDZIJA, Nikola
• Granted Patent: US 8,014,880 B2
• Time in Patent Office: Days
• US Class Current: 700/109
• CPC Class Codes: G05B 23/021 adopting a different treatment of each operating region or a different mode of the monitored system, e.g. transient modes; different operating configurations of monitored system
• 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 23/0245 based on a qualitative model, e.g. rule based; if-then decisions
• CPC Class Codes: Y10S 715/965 for process control and configuration

Abstract

A method and system for monitoring a process in a process plant includes collecting data from a process control system within the process plant, where the data is representative of a normal operation of the process when the process is on-line and operating normally, performing a multivariate statistical analysis to represent the normal operation of the process based on a set of collected on-line process data comprising a measure of the normal operation of the process when the process is on-line and operating normally, and representing a real-time on-line operation of the process using monitored on-line process data comprising a measure of a real-time operation of the process when the process is on-line as an input to the representation of the normal operation of the process.

Claims

1. A system for monitoring a process in a process plant, the system comprising:

a data collection tool adapted to collect on-line process data from a process control system within the process plant, wherein the collected on-line process data is representative of a normal operation of the process when the process is on-line and operating normally; and

an analysis tool comprising a multivariate statistical analysis engine adapted to represent the normal operation of the process based on a set of collected on-line process data comprising a measure of the normal operation of the process when the process is on-line and operating normally, and wherein the analysis tool is adapted to represent a real-time on-line operation of the process using monitored on-line process data comprising a measure of a real-time operation of the process when the process is on-line as an input to the representation of the normal operation of the process.

2. The system of claim 1, wherein the data collection tool is adapted to collect the on-line process data from the process control system based on a user-defined set of input variables within the process, wherein the collected on-line process data comprises a plurality of observations of each of the input variables.

3. The system of claim 1, wherein the data collection tool is adapted to store the collected on-line process data from the process control system.

4. The system of claim 1, wherein the analysis tool is implemented within the process control system of the process plant.

5. The system of claim 1, wherein the collected on-line process data comprises multivariable process data from the process and representative of the normal operation of the process when the process is on-line and operating normally.

6. The system of claim 1, wherein the monitored on-line process data comprises multivariable process data from the process and representative of a real-time on-line operation of the process.

7. The system of claim 1, wherein the collected on-line process data comprises raw process variable data from the process and representative of the normal operation of the process when the process is on-line and operating normally.

8. The system of claim 1, wherein the monitored on-time process data comprises raw process variable data from the process and representative of a real-time on-line operation of the process.

9. The system of claim 1, wherein the analysis tool is adapted to perform a statistical analysis on the collected on-line process data to determine a threshold value.

10. The system of claim 9, wherein the analysis tool is adapted to generate an output from the representation of the normal operation of the process based on the input of the monitored on-line process data to the representation of the normal operation of the process, and wherein the analysis tool is adapted to generate an alarm if the output from the representation of the normal operation of the process exceeds the threshold value.

11. The system of claim 9, wherein the analysis tool is adapted to generate an output from the representation of the normal operation of the process based on the input of the monitored on-line process data to the representation of the normal operation of the process, and wherein the analysis tool is adapted to determine the presence of an abnormal process condition if the output from the representation of the normal operation of the process exceeds the threshold value.

12. The system of claim 11, wherein the analysis tool is adapted to generate an alarm based on the presence of an abnormal process condition.

13. The system of claim 9, wherein the analysis tool is adapted to determine the threshold value based upon a user-defined confidence level.

14. The system of claim 1, wherein the multivariate statistical analysis engine is adapted to perform a principal component analysis on the set of collected on-line process data to represent the normal operation of the process in a loading matrix defining a subspace.

15. The system of claim 14, wherein the analysis tool is adapted to project the monitored on-line process data into the subspace defined by a loading matrix generated by the multivariate statistical engine which represents the normal operation of the process.

16. The system of claim 1, wherein the multivariate statistical analysis engine is adapted to represent the normal operation of the process in each of a plurality of on-line states based on different sets of collected on-line process data, wherein each set of collected on-line process data comprises a measure of the normal operation of the process when the process is on-line and operating normally in one of the plurality of on-time states.

17. The system of claim 16, wherein the analysis tool is adapted to represent a real-time on-line operation of the process using monitored on-line process data comprising a measure of a real-time operation of the process when the process is on-line and operating in one of the plurality of on-line states as an input to the representation of the normal operation of the process corresponding to the corresponding one of the plurality of on-line states.

18. The system of claim 17 wherein the analysis tool is adapted to use a representation of the normal operation of the process in one of the plurality of on-line states selected by a user from one of the plurality of representations of the normal operation of the process.

19. The system of claim 1, wherein the multivariate statistical analysis engine is adapted to represent the normal operation of the process in a user-selected on-line state based on a set of collected on-line process data comprising a measure of the normal operation of the process when the process is on-line and operating normally in the user-selected on-line state, and wherein the analysis tool is adapted to represent a real-time on-line operation of the process using monitored on-line process data comprising a measure of a real-time operation of the process when the process is on-line as an input to the representation of the normal operation of the process in the user-selected on-line state.

20. The system of claim 1 wherein the analysis tool is adapted to generate a plurality of outputs from the representation of the normal operation of the process based on the input of the monitored on-line process data to the representation of the normal operation of the process, the system further comprising a graphical interface tool adapted to generate a visualization of at least one of the plurality of outputs from the representation of the normal operation of the process.

21. The system of claim 20, wherein the graphical interface tool is implemented within an interface of the process control system.

22. The system of claim 20, wherein the visualization comprises a plot of data of one of the plurality of outputs from the representation of the normal operation of the process versus time.

23. The system of claim 20) wherein the visualization comprises a plot of data of two or more of the plurality of outputs from the representation of the normal operation of the process, wherein each of the two or more outputs comprise a reference coordinate for the plot.

24. As method for monitoring a process in a process plant, the method comprising:

collecting data from a process control system within the process plant, wherein the data is representative of a normal operation of the process when the process is on-line and operating normally;

performing a multivariate statistical analysis to represent the normal operation of the process based on a set of collected on-line process data comprising a measure of the normal operation of the process when the process is on-line and operating normally; and

representing a real-time on-line operation of the process using monitored on-line process data comprising a measure of a real-time operation of the process when the process is on-line as an input to the representation of the normal operation of the process.

25. The method of claim 24, wherein collecting data from the process control system comprises collecting data from the process control system based on a user-defined set of input variables within the process wherein the collected data comprises a plurality of observations of each of the input variables.

26. The method of claim 24, further comprising storing the collected data from the process control system.

27. The method of claim 24, wherein the collected data comprises multivariable process data from the process and representative of the normal operation of the process when the process is on-line and operating normally.

28. The method of claim 24, wherein the monitored on-line process data comprises multivariable process data from the process and representative of a real-time on-line operation of the process.

29. The method of claim 24, wherein the collected data comprises raw process variable data from the process and representative of the normal operation of the process when the process is on-line and operating normally.

30. The method of claim 24, wherein the monitored on-line process data comprises raw process variable data from the process and representative of a real-time on-line operation of the process.

31. The method of claim 24, further comprising performing a statistical analysis on the collected data to determine a threshold value.

32. The method of claim 31, further comprising:

generating an output from the representation of the normal operation of the process based on the input of the monitored on-line process data to the representation of the normal operation of the process; and

generating an alarm if the output from the representation of the normal operation of the process exceeds the threshold value.

33. The method of claim 31, further comprising:

generating an output from the representation of the normal operation of the process based on the input of the monitored on-line process data to the representation of the normal operation of the process; and

determining the presence of an abnormal process condition if the output from the representation of the normal operation of the process exceeds the threshold value.

34. The method of claim 33, further comprising generating an alarm based on the presence of an abnormal process condition.

35. The method of claim 31, further comprising determining the threshold value based upon a user-defined confidence level.

36. The method of claim 24, wherein performing a multivariate statistical analysis to represent the normal operation of the process comprises performing a principal component analysis on the set of collected on-line process data to represent the normal operation of the process in a loading matrix defining a subspace.

37. The method of claim 36, further comprising projecting the monitored on-line process data into the subspace defined by the loading matrix.

38. The method of claim 24, wherein performing a multivariate statistical analysis to represent the normal operation of the process comprises performing a plurality of multivariate statistical analyses, wherein performing each multivariate statistical analysis generates a representation of the normal operation of the process in a different on-line state based on different sets of collected data, wherein each set of collected data comprises a measure of the normal operation of the process when the process is on-line and operating normally in one of the plurality of on-line states.

39. The method of claim 38, wherein representing a real-time on-line operation of the process using monitored on-line process data comprises representing a real-time on-line operation of the process using monitored on-line process data comprising a measure of a real-time operation of the process when the process is on-line and operating in one of the plurality of on-line states as an input to the representation of the normal operation of the process corresponding to the corresponding one of the plurality of on-line states.

40. The method of claim 39, further comprising using a representation of the normal operation of the process in one of the plurality of on-line states selected by a user from one of the plurality of representations of the normal operation of the process.

41. The method of claim 24, wherein representing a real-time on-line operation of the process using monitored on-line process data comprises representing the normal operation of the process in a user-selected on-line state based on a set of collected data comprising a measure of the normal operation of the process when the process is on-line and operating normally in the user-selected on-line state, and wherein representing a real-time on-line operation of the process using, monitored on-line process data comprises representing a real-time on-line operation of the process using monitored on-line process data comprising a measure of a real-time operation of the process when the process is on-line as an input to the representation of the normal operation of the process in the user-selected on-line state.

42. The method of claim 24, further comprising:

generating a plurality of outputs from the representation of the normal operation of the process based on the input of the monitored on-line process data to the representation of the normal operation of the process; and

generating a visualization of at least one of the plurality of outputs from the representation of the normal operation of the process.

43. The method of claim 42, wherein the visualization comprises a plot of data of one of the plurality of outputs from the representation of the normal operation of the process versus time.

44. The method of claim 42, wherein the visualization comprises a plot of data of two or more of the plurality of outputs from the representation of the normal operation of the process, wherein each of the two or more outputs comprise a reference coordinate for the plot.

45. A process control system for monitoring a process in a process plant comprising:

an analysis application stored on a computer readable memory and adapted to be executed on a processor to generate a multivariate statistical representation of a normal on-line operation of the process based on a set of collected on-line process data collected from the process, wherein the collected on-line process data comprises a higher-dimensional space of observations for each of a plurality of process variables within the process when the process is on-line and operating normally, and the multivariate statistical representation of the normal operation of the process represents the normal operation of the process in a lower-dimensional space as defined by significant variations in the process, and wherein the analysis tool is adapted to project monitored on-line process data onto the lower-dimensional space to generate one or more scores of the multivariate statistical representation, wherein the monitored on-line process data comprises a higher-dimensional space of real-time on-line observations for each of the plurality of process variables within the process when the process is on-line, and the one or more scores of the multivariate statistical representation represent the real-time on-line operation of the process in the lower-dimensional space as a function of the significant variations in the process; and

a display application stored on a computer readable memory and adapted to be executed on a processor to generate a real-time visualization of at least one of the one or more scores of the multivariate statistical representation.

46. The process control system of claim 45, wherein the analysis application is adapted to be executed on a processor to generate the multivariate statistical representation using principal component analysis.

47. The process control system of claim 45, wherein the analysis application is adapted to be executed on a processor to determine the presence of an abnormal process condition if a score of the multivariate statistical representation exceeds a threshold value.

48. The process control system of claim 47, wherein the analysis application is adapted to be executed on a processor to generate an alarm based on the presence of an abnormal process condition.

49. The process control system of claim 47, wherein the display application is adapted to be executed on a processor to generate the real-time visualization of the at least one of the one or more scores of the multivariate statistical representation with a visualization of the threshold value.

50. The process control system of claim 45, wherein the display application is adapted to be executed on a processor to generate a real-time visualization of at least one of the one or more scores of the multivariate statistical representation as a function of time.

51. The process control system of claim 45, wherein the display application is adapted to be executed on a processor to generate a real-time visualization of at least one of the one or more scores of the multivariate statistical representation as a function of another score of the multivariate statistical representation.