Method and apparatus for managing process transitions

• US 20050114078A1
• Filed: 11/02/2004
• Published: 05/26/2005
• Est. Priority Date: 11/14/2000
• Status: Active Grant

• Current Assignee: National University of Singapore
• Original Assignee: National University of Singapore
• Inventors: Viswanathan, Pradeep Kumar, Sundarraman, Anandakrishnan, Srinivasan, Rajagopalan, Vedam, Hiranmayee, Ananthanarayanan, Nochur
• Granted Patent: US 7,233,882 B2
• Time in Patent Office: Days
• US Class Current: 702/179
• 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 23/0264 Control of logging system, e.g. decision on which data to store; time-stamping measurements

Abstract

Operators and plant computer programs used to operate a complex process facility are aided in managing the process during transitions by a computer-based apparatus. The apparatus incorporates a knowledge base and methods for identifying modes and transitions during plant operation. At frequent intervals, measured values of the process variables are used to evaluate the current state of the process and its sections and subsections. The identified state of the plant is broadcast to different clients of this application. The apparatus monitors the plant for the normal execution of the transition. It also identifies the current task being performed in the process and sends this message to different sections of the plant. The results are displayed on a visual display device and can also be sent to other plant computer programs for guidance during the transition. A permanent chronological record of the sequence of events—modes and transitions—of the plant and sections and subsections including the pertinent plant conditions and information is also generated by the apparatus for subsequent review and analysis. Methods for generating the knowledge base are also presented.

Claims

1. A system for monitoring a complex process over a transition phase from a first mode, said system comprising:

a database containing identifying data related to known modes and known transitions;

a measurement device providing current process data related to said complex process;

a comparator using said identifying data and said current process data so as to determine which one of said known modes and said known transitions that said complex process is currently within; and

a range checking component using stored data of said one of said known modes and said known transitions as determined by said comparator, and using said current process as data so as to signal when said complex process is outside of an acceptable range.

2. A system according to claim 1 wherein said transition phase includes at least one changing process variable.

3. A system according to claim 1 wherein said mode includes all substantially constant variables.

4. A system according to claim 1 wherein said complex process comprises a second mode after said transition phase.

5. A system according to claim 1 wherein said identifying data related to said known modes and said known transitions is based on historical data.

6. A system according to claim 1 wherein said identifying data related to said known transitions are based on theoretical data.

7. A system according to claim 1 wherein said comparator is configurable to receive input from an operator to determine whether said complex process is currently within one of said known modes and said known transitions.

8. A system according to claim 1 wherein each of said known modes and said known transitions is subdivided into enhanced states, and further wherein said database contains identifying data related to each of said enhanced states of said known modes and said known transitions so as to determine which one of said enhanced states of said known modes and said known transitions that said complex process is currently within.

9. A system according to claim 8 wherein each of said enhanced states is subdivided into atoms, and further wherein said database contains identifying data related to each of said atoms of said enhanced states of said known modes and said known transitions so as to determine which one of said atoms of said enhanced states of said known modes and said known transitions that said complex process is currently within.

10. A system according to claim 1 wherein said identifying data of said comparator comprises mapped data.

11. A system according to claim 10 wherein at least a portion of said mapped data relates to said known transitions.

12. A system according to claim 10 wherein at least a portion of said mapped data relates to known modes.

13. A system according to claim 10 wherein at least a portion of said mapped data relates to an individual subsection of said known transitions.

14. A system according to claim 10 wherein at least a portion of said mapped data relates to an individual subsection of said known modes.

27. A system according to claim 1 wherein said identifying data related to said known transitions are based on experience-based data.

15. A method for monitoring a complex process over a transition phase from a first mode, said method comprising:

providing a database containing identifying data related to known modes and known transitions;

obtaining current process data related to said complex process;

comparing said current process data with said identifying data so as to determine which one of said known modes and said known transitions that said complex process is currently within; and

comparing said current data with stored data of said one of said known modes and said known transitions as determined by said comparator so as to signal when said complex process is outside of an acceptable range.

16. A method for supervising a complex process involving a client, said method comprising:

identifying a state of said complex process, said state comprising one of at least one transition phase and at least one mode;

reporting said identified state comprising said one of at least one transition phase and at least one mode to said client; and

reconfiguring said client based on said identified state of state of said complex process.

17. A method for classifying a complex process into modes and transition phases, said method comprising:

identifying values of given variables within a subsection of said complex process;

determining the magnitude of variation of each of said values of said given variables over said subsection of said complex process; and

classifying said subsection into one of said modes and said transition phases based on said magnitude of variation of each of said values of said given variables over said subsection of said complex process.

18. A method according to claim 17 wherein a section is classified in a mode when each of its subsections are classified in modes.

19. A method according to claim 17 wherein a section is classified in a transition when at least one of its subsections is classified in a transition phase.

20. A method of generating knowledge base for monitoring a complex process over a transition, said method comprising:

obtaining plant history data and plant hierarchy;

grouping plant tags into sections and subsections;

generating a process trend and a wavelet coefficient for each said plant tags;

generating process trends for each tag in every lower-level group;

identifying periods of constancy for each variable in each lower-level group using said generated process trends;

identifying modes for each lower-level group;

consolidating and characterizing modes for each lower-level group;

consolidating and characterizing transitions;

generating an event list for each lower-level group;

generating an event lookup table for higher-level groups; and

generating said knowledge base.

21. A method for identifying a state of a complex process, said state comprising one of at least one transition phase and at least one mode, said method comprising:

providing tagged sensor data from said complex process;

segregating said tagged sensor data into plant sections and subsections of a plant hierarchy using a knowledge base;

analyzing the range of variables of each of said subsections using said knowledge base so as to determine which one of a transition phase with at least one of said variables outside of said pre-determined range and a mode with all of said variables within said pre-determined phase that said subsection is currently within;

identifying a current mode for each of said subsections in one of said modes based on values of said variables and said knowledge base;

identifying current transition phases for each of said subsections in one of said transition phases using wavelet-based feature comparison for key variables with said knowledge base; and

deducing events in higher level plant groups using each of said current modes, current transitions, and said knowledge base.

22. A method for identifying a current transition phase of a complex process, said method comprising:

generating wavelet approximation coefficients for variables in a subsection in said current transition phase;

comparing said wavelet approximation coefficients of said current transition phase with corresponding wavelet coefficients for transitions available in a knowledge base; and

identifying said current transition phase based on said comparison of said wavelet approximation coefficients of said current transition phase with said corresponding wavelet coefficient for transitions available in said knowledge base.

23. A method to track a complex process within transition phase, said method comprising:

providing distinct features of process variables of common operation data within a common transition phase;

measuring distinct features of process variables of said complex process within said transition phase; and

comparing said distinct features of said process variables of said complex process within said transition phase and said distinct features of said process variables of said common operation data within said common transition phase so as to track said complex process.

24. A method for determining an enhanced trend for a series of one subset of distinct features of a process variable, said method comprising:

obtaining stored information related to said series of said one subset of said distinct features of said process variable;

obtaining current information related to said one subset of said distinct features of said process variable; and

comparing said current information related to said one subset of said distinct features of said process variable and said stored information related to said series of said one subset of said distinct features of said process variable to one another so as to determine said enhanced trend for said series of said one subset of distinct features of said process variable.

25. A method for tracking a trend of a process variable against a stored trend as given in a dictionary, said method comprising:

comparing said trend of said process variable to a current dictionary atom and a subsequent dictionary atom;

identifying one of said current dictionary atom and said subsequent dictionary atom corresponding to said trend of said process variable; and

broadcasting information related to current transition phase of said identified current dictionary atom to sections of a plant.

26. A method for monitoring a complex process during a transition phase, said method comprising:

measuring a real-time enhanced trend of process variables;

identifying a dictionary enhanced trend of said process variables;

comparing trends of each of said real-time enhanced trend and said dictionary enhanced trend to one another;

comparing process variable magnitudes of each of said real-time enhanced trend of said process variables and said dictionary enhanced trend to one another;

comparing trend durations of each of said real-time enhanced trend of said process variables and said dictionary enhanced trend to one another;

calculating a matching degree for each of said process variables from said trend comparison, said process variable magnitude comparison, and said trend duration comparison;

calculating a degree of fault of said complex process as a function of degree of fault of individual sensors; and

informing an operator of said matching degree and said degree of fault.