MULTIVARIATE DETECTION OF TRANSIENT REGIONS IN A PROCESS CONTROL SYSTEM

• US 20080091390A1
• Filed: 09/28/2007
• Published: 04/17/2008
• 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: Hamad, Ahmad, SAMARDZIJA, Nikola
• Granted Patent: US 7,966,149 B2
• Time in Patent Office: Days
• US Class Current: 703/2
• 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

Methods and systems to detect transient operations from abnormal operations, and to detect abnormal operations in a coker heater, include collecting on-line process data. The collected on-line process data is generated from a plurality of process variables of the process, or coker heater. A first representation of the operation of the process, or coker heater, is generated based on a first set of the collected on-line process data generated from a first set of the process variables. The first representation is adapted to be executed to generate a first result. A second representation of the operation of the process, or coker heater, is generated based on the first result and based on a second set of the collected on-line process data generated from a second set of the process variables. The second representation is adapted to be executed to generate a prediction of data generated from the second set of the process variables. The prediction is analyzed to detect an abnormal operation or to detect whether one or more abnormal operations comprises a transient operation of the process.

Claims

1. A system for facilitating detection and identification of a transient operation from an abnormal operation of 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 comprises data representative of an operation of the process when the process is on-line and wherein the collected on-line process data is generated from a plurality of process variables of the process;

a first analysis tool comprising a first multivariate statistical analysis engine adapted to generate a first representation of the operation of the process based on a first set of the collected on-line process data generated from a first set of the process variables of the process, wherein the first representation of the operation of the process is adapted to be executed to generate a first result;

a second analysis tool comprising a second multivariate statistical analysis engine adapted to generate a second representation of the operation of the process based on the first result and based on a second set of the collected on-line process data generated from a second set of the process variables of the process, wherein the second representation of the operation of the process is adapted to be executed to generate a prediction of data generated from the second set of the process variables; and

a monitoring tool adapted to analyze the prediction to detect whether one or more abnormal operations detected based on the first representation of the operation of the process comprises a transient operation of the process.

2. The system of claim 1, wherein the first analysis tool is adapted to represent the on-line operation of the process using monitored on-line process data comprising a measure of the operation of the process when the process is on-line as an input to the first representation of the operation of the process

3. The system of claim 1, wherein the first multivariate statistical analysis engine is adapted to generate an output from the first representation of the operation of the process based on an input of monitored on-line process data to the first representation of the operation of the process, and wherein the monitoring tool is adapted to determine the presence of an abnormal process condition based on the output from the first representation of the normal operation of the process.

4. The system of claim 1, wherein the first multivariate statistical analysis engine is adapted to perform a principal component analysis on the first set of the collected on-line process data to generate the first representation of the operation of the process, and wherein the second multivariate statistical analysis engine is adapted to perform at least one of a principal component regression analysis or a partial least squares analysis on the second set of the collected on-line process data to generate a prediction of data generated from the second set of the process variables as a function of the first result.

5. The system of claim 4, wherein the first result comprises one or more components demonstrating the greatest variance within the first set of the collected on-line process data, and wherein the second multivariate statistical analysis engine is adapted to generate the prediction of data generated from the second set of the process variables as a function of the one or more components.

6. The system of claim 1, wherein the first set of the collected on-line process data comprises a set of the on-line process data generated from a plurality of independent process variables of the process, and wherein the second set of collected on-line process data comprises a set of the on-line process data generated from a dependent process variable of the process.

7. The system of claim 1, wherein the monitoring tool is adapted to prevent an abnormal condition alert from generating if a detected abnormal operation comprises a transient operation of the process.

8. The system of claim 1, wherein the monitoring tool is adapted to estimate a transient time constant associated with a transient condition.

9. The system of claim 1, further comprising a display tool adapted to generate a visualization comprising a first plot of a result generated by the first representation of the operation of the process and a second plot of a result generated by the second representation of the operation of the process.

10. A method of identifying a transient operation from an abnormal operation of a process in a process plant, the method comprising:

collecting on-line process data from a process control system within the process plant, wherein the data is representative of an operation of the process when the process is on-line, and wherein the collected on-line process data is generated from a plurality of process variables of the process comprising a first data space having a plurality of dimensions;

generating a first model of the operation of the process using a first set of the collected on-line process data generated from a first set of the process variables of the process, the first model comprising a measure of the operation of the process when the process is on-line within a second data space having fewer dimensions than the first data space;

generating a second model of the operation of the process using components of the second data space and using a second set of the collected on-line process data generated from a second set of the process variables of the process;

determining the presence of an abnormal operation based on an output from the first model, wherein the model is executed based on an input of monitored on-line process data to the first model of the operation of the process;

generating a prediction of data generated from the second set of the process variables as a function of the components of the second data space using the second model; and

determining if the abnormal operation comprises a transient condition based on the prediction of the data generated from the second set of the process variables representing the occurrence of the transient condition corresponding to the occurrence of the abnormal operation.

11. The method of claim 10, wherein generating a first model comprises performing a multivariate statistical analysis to represent the operation of the process based on the first set of the collected on-line process data generated from the first set of the process variables of the process.

12. The method of claim 11, wherein performing a multivariate statistical analysis comprises performing a principal component analysis on the first set of the collected on-line process data to represent the operation of the process in a loading matrix defining a subspace.

13. The method of claim 12, wherein determining the presence of an abnormal operation comprises projecting monitored on-line process data generated from the first set of the process variables of the process into the subspace defined by the loading matrix.

14. The method of claim 10, wherein generating a second model comprises performing a multivariate statistical analysis to represent the operation of the process based on the second set of the collected on-line process data generated from the second set of the process variables of the process.

15. The method of claim 14, wherein performing a multivariate statistical analysis comprises performing at least one of a principal component regression analysis or a partial least squares analysis to predict the second set of the process variables as a function of the components of the second data space.

16. A method of detecting an abnormal operation of a coker heater, comprising:

collecting on-line process data for the coker heater, wherein the collected on-line process data is representative of an operation of the coker heater when the coker heater is on-line, and wherein the collected on-line process data is generated from a plurality of process variables of the coker heater;

performing a first multivariate statistical analysis to represent the normal operation of the coker heater in a known state based on a first set of the collected on-line process data generated from a first set of the process variables of the coker heater comprising a multivariate data structure, the representation comprising a measure of the operation of the coker heater in a data space having fewer dimensions than the multivariate data structure;

performing a second multivariate statistical analysis to represent the operation of the coker heater based on a second set of the collected on-line process data generated from a second set of the process variables of the coker heater as a function of components from the data space having fewer dimension than the multivariate data structure to output a prediction of data generated from the second set of the process variables of the coker heater; and

determining if a corresponding signal generated from the second set of the process variables of the coker heater deviates from the prediction of the data generated from the second set of the process variables of the coker heater to detect an abnormal situation within the coker heater.

17. The method of claim 16, wherein performing a first multivariate statistical analysis comprises performing a principal component analysis on the first set of the collected on-line process data to represent the operation of the process in a loading matrix defining a subspace.

18. The method of claim 16, wherein performing a second multivariate statistical analysis comprises performing at least one of a principal component regression analysis or a partial least squares analysis to predict the second set of the process variables as a function of the components from the data space having fewer dimension than the multivariate data structure.

19. The method of claim 16, wherein the abnormal condition comprises coking in a tube pass of the coker heater.

20. The method of claim 16, wherein the collected on-line process data is representative of a normal operation of the coker heater when the coker heater is on-line and operating normally.

21. The method of claim 16, wherein the first set of the process variables of the coker heater comprises one or more independent variables selected from the group consisting of:

a temperature of pass matter at a position before a heating element of a conduit of the coker heater, a flow rate of each pass within the coker heater, a total flow rate of pass matter, a temperature of pass matter at a position after the heating element of the conduit of the coker heater, a flow rate of product material from the coker heater, and a final temperature of product material from the coker heater, and

wherein the second set of the process variables of the coker heater comprises one or more dependent variables including an exit temperature of a pass within the coker heater.

22. A system for detecting an abnormal operation in a coker heater in a process plant, comprising:

a model of the normal operation of the coker heater based on a set of collected on-line process data comprising a measure of the normal operation of the coker heater when the coker heater is on-line and operating normally, the model including a first multivariate statistical model based on a first set of the collected on-line process data generated from a set of independent process variables of the coker heater having a first data space having a plurality of dimensions, wherein the first multivariate statistical model represents the operation of the operation of the coker heater in a loading matrix defining a subspace, and wherein the model includes a second multivariate statistical model based on a second set of the collected on-line process data generated from a set of dependent variables of the coker heater and based on one or more components from the loading matrix; and

a deviation detector coupled to the model, the deviation detector configured to determine if a dependent variable value from the set of dependent variables significantly deviates from a predicted dependent variable value generated from the second multivariate statistical model by comparing a difference between the dependent variable value and predicted dependent variable value.

23. The system of claim 22, wherein the first multivariate statistical model comprises a model generated from a principal component analysis on the first set of the collected on-line process data, and wherein the second multivariate statistical model comprises a model generated from at least one of a principal component regression analysis or a partial least squares analysis on the second set of the collected on-line process data.

24. The system of claim 22, wherein a significant deviation of the dependent variable value from the predicted dependent variable value comprises coking in a tube pass of the coker heater.

25. The system of claim 22, further comprising a display tool adapted to generate a visualization comprising a first plot of a result generated by the representation of the operation of the coker heater from the first multivariate statistical model and a second plot of a result generated by a representation of the operation of the coker heater from the second multivariate statistical model.