Method and System for Detecting Faults in a Process Plant

• US 20080188972A1
• Filed: 10/11/2006
• Published: 08/07/2008
• Est. Priority Date: 10/11/2006
• Status: Abandoned Application

• Current Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Original Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Inventors: Miller, John P.
• Application Number:
  US11/548,590
• Publication Number:
  US 20080188972A1
• Time in Patent Office: Days
• US Class Current: 700/110
• CPC Class Codes: G05B 19/4184 characterised by fault tolerance, reliability of production system
• CPC Class Codes: G05B 2219/31357 Observer based fault detection, use model
• CPC Class Codes: G05B 2219/32224 Identify parameters with highest probability of failure
• 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 23/0254 based on a quantitative model, e.g. mathematical relationships between inputs and outputs; functions: observer, Kalman filter, residual calculation, Neural Networks
• CPC Class Codes: G06F 18/2433 Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
• CPC Class Codes: G06F 2218/08 Feature extraction
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
• CPC Class Codes: Y02P 90/80 Management or planning

Abstract

In methods and systems that may facilitate detecting abnormal operation in a process plant, values of a plurality of process variables may be analyzed to determine whether any of a plurality of faults associated with the process plant exist. If one or more faults are detected, one or more indicators may be generated. Analyzing the values of the plurality of process variables may include utilizing a coefficient matrix. The coefficient matrix may be generated based on process variable data corresponding to the known occurrences of faults.

Claims

1. A method for facilitating detection of abnormal operation of a process in a process plant, comprising:

receiving first data corresponding to a plurality of process variables of a process plant;

using the first data and a coefficient matrix to generate a fault observation vector; and

determining whether there is abnormal operation of the process based on the fault observation vector.

2. A method according to claim 1, further comprising creating a process variable vector using the received first data;

wherein using the first data and the coefficient matrix to generate a fault observation vector comprises multiplying the process variable vector and the coefficient matrix.

3. A method according to claim 2, further comprising creating a process variable matrix using the received first data, the process variable matrix including the process variable vector;

wherein using the first data and the coefficient matrix to generate a fault observation vector comprises multiplying the process variable matrix and the coefficient matrix to generate a fault observation matrix including the fault observation vector.

4. A method according to claim 1, wherein determining whether there is abnormal operation of the process comprises determining whether there are one or more faults of a plurality of faults based on the fault observation vector.

5. A method according to claim 4, wherein determining whether there are one or more faults comprises comparing components of the fault observation vector to a threshold.

6. A method according to claim 5, further comprising generating an indicator of one fault from the plurality of faults if a component of the fault observation vector, corresponding to the one fault, exceeds a threshold.

7. A method according to claim 1, wherein determining whether there is abnormal operation of the process comprises providing components of the fault observation vector to at least one of an expert system, a neural network system, or a fuzzy logic system.

8. A method according to claim 1, further comprising:

receiving second data corresponding to the plurality of process variables, the second data corresponding to the occurrence of a plurality of faults;

generating the coefficient matrix using the second data.

9. A method according to claim 8, further comprising:

receiving third data corresponding to the plurality of process variables, the third data corresponding to the nonoccurrence of any faults of the plurality of faults;

wherein generating the coefficient matrix comprises generating the coefficient matrix using the third data.

10. A tangible medium storing machine readable instructions, the machine readable instructions capable of causing one or more machines to:

receive data corresponding to a plurality of process variables of a process plant;

use the first data and a coefficient matrix to generate a fault observation vector; and

determine whether there is abnormal operation of the process based on the fault observation vector.

11. A system for facilitating detection of abnormal operation of a process in a process plant, comprising:

a fault observation vector generator coupled to receive a coefficient matrix and process variable data associated with the process plant; and

an abnormal operation detection system, coupled to the fault observation vector generator, to detect abnormal operation of the process plant based on a fault observation vector.

12. A system according to claim 11, wherein the abnormal operation detection system is configured to compare components of the fault observation vector to a threshold and to generate indicators of faults based on the comparison.

13. A system according to claim 11, wherein the abnormal operation detection system comprises at least one of an expert system, a neural network system, or a fuzzy logic system.

14. A system according to claim 11, further comprising a coefficient matrix generator coupled to the fault observation vector generator, the coefficient matrix generator configured to generate the coefficient matrix based on process variable data corresponding to occurrences of faults of a plurality of faults.

15. A system according to claim 14, wherein the coefficient matrix generator is configured to generate the coefficient matrix further based on process variable data corresponding to non-occurrences of any of the faults of the plurality of faults.

16. A system according to claim 14, wherein the coefficient matrix generator is implemented in a first device of the process plant, wherein the fault observation vector generator is implemented in a second device of the process plant, and wherein the abnormal operation detection system is implemented in at least a third device of the process plant.

17. A system according to claim 14, wherein the coefficient matrix generator and the fault observation vector generator are implemented in a first device of the process plant, and wherein the abnormal operation detection system is implemented in at least a second device of the process plant.

18. A system according to claim 14, wherein the coefficient matrix generator, the fault observation vector generator, and the abnormal operation detection system are implemented in a single device of the process plant.

19. A method for configuring an abnormal operation detection system for a process plant, comprising:

receiving first process variable data corresponding to the occurrences of faults of a plurality of faults of a process system in the process plant;

generating a process variable data matrix based on the first process variable data;

generating a fault matrix corresponding to the process variable data matrix; and

generating a coefficient matrix using the process variable data matrix and the fault matrix, the coefficient matrix to be used by an abnormal operation detection system to generate indicators of faults based on process variable data received by the abnormal operation detection system.

20. A method according to claim 19, further comprising receiving second process variable data corresponding to the non-occurrence of any faults of the plurality of faults;

wherein generating the process variable data matrix comprises generating the process variable data matrix further based on the second process variable data.

21. A method according to claim 20, wherein the first process variable data includes process variable data corresponding to individual occurrences of each fault of the plurality of faults.

22. A method according to claim 19, wherein generating the process variable data matrix comprises including bias terms in the process variable data matrix.

23. A method according to claim 19, wherein generating the coefficient matrix comprises generating the coefficient matrix according to a regression technique.

24. A method according to claim 23, wherein generating the coefficient matrix comprises generating the coefficient matrix according to the equation:

A=(X^TX)<sup>−

1</sup>X^TF;

wherein A is the coefficient matrix, X is the process variable data matrix, and F is the fault matrix.

25. A method according to claim 23, wherein each row of the process variable data matrix X corresponds to different set of the first process variable data;

wherein each column of the fault matrix F corresponds to a different fault from the plurality of faults;

wherein each row of the fault matrix F corresponds to a different row of the process variable data matrix;

wherein generating the fault matrix comprises, for each row of the fault matrix F, inserting a non-zero value in the row at a column that corresponds to the fault, if any, associated with the corresponding row of the process variable data matrix X and inserting zero values in the remaining columns.

26. A method according to claim 25, wherein the non-zero value is one.

27. A method according to claim 23, wherein generating the process variable data matrix comprises creating a column having a bias value in each row.

28. A method according to claim 27, wherein the bias value is one.

29. A method according to claim 23, wherein generating the coefficient matrix comprises generating the coefficient matrix according to at least one of an ordinary least squares (OLS) technique, a multiple linear regression (MLR) technique, a partial least squares (PLS) technique, a principal component analysis (PCA) technique, a principal component regression (PCR) technique, a ridge regression (RR) technique, a variable subset selection (VSS) technique, or a support vector machines (SVM) technique.

30. A method according to claim 23, wherein generating the coefficient matrix comprises generating the coefficient matrix according to a non-linear regression technique.

31. A method according to claim 30, wherein the non-linear regression technique comprises at least one of using higher-order powers of the process variables, using cross-terms of the process variables, using nonlinear functions of the process variables, or a neural network technique.

32. A method according to claim 23, wherein generating the coefficient matrix comprises generating the coefficient matrix according to a regression technique utilizing time-delayed values of elements of the process variable data matrix.

33. A method according to claim 19, further comprising transmitting the coefficient matrix over a communication link in the process plant.

34. A method according to claim 33, wherein transmitting the coefficient matrix comprises transmitting the coefficient matrix to a field device over a bus.

35. A tangible medium storing machine readable instructions, the machine readable instructions capable of causing one or more machines to:

receive process variable data corresponding to the occurrences of faults of a plurality of faults of a process system in the process plant;

generate a process variable data matrix based on the process variable data;

generate a fault matrix corresponding to the process variable data matrix; and

generate a coefficient matrix using the process variable data matrix and the fault matrix, the coefficient matrix to be used by an abnormal operation detection system to generate indicators of faults based on process variable data received by the abnormal operation detection system.

36. A system for facilitating detection of abnormal operation of a process in a process plant, comprising:

at least one computer readable medium;

at least one processor coupled to the at least one computer readable medium, the processor configured according to executable instructions stored on the at least one computer readable medium to;

receive process variable data corresponding to the occurrences of faults of a plurality of faults of a process system in the process plant;

generate a process variable data matrix based on the process variable data;

generate a fault matrix corresponding to the process variable data matrix; and

generate a coefficient matrix using the process variable data matrix and the fault matrix, the coefficient matrix to be used by an abnormal operation detection system to generate indicators of faults based on process variable data received by the abnormal operation detection system.