Process facility monitor using fuzzy logic

• US 5,351,200 A
• Filed: 11/22/1991
• Issued: 09/27/1994
• Est. Priority Date: 11/22/1991
• Status: Expired due to Term

• Current Assignee: Westinghouse Electric Company LLC (Brookfield Asset Management Incorporated)
• Original Assignee: Westinghouse Electric Company LLC (Brookfield Asset Management Incorporated)
• Inventors: Impink, Albert J. Jr.
• Primary Examiner(s): Cosimano, Edward R.
• Application Number:
  US07/796,291
• Time in Patent Office: 1,040 Days
• Field of Search: 364/130, 364/550, 395/900, 395/906, 395/907
• US Class Current: 702/181
• CPC Class Codes: G05B 23/0229 knowledge based, e.g. expert systems; genetic algorithms
• CPC Class Codes: Y10S 706/90 Fuzzy logic

Abstract

A method for monitoring operation of a process facility comprising the steps of storing at least one planned sequence of hypothesized plant states using multi-valued logic to define calculation of a congruence factor for each of the hypothesized plant states; wherein said storing step includes storing relationships between the multi-valued logic variables defined by a maximum value function when any of the multi-valued logic variables in a first relationship may be true and defined by an average function when all of the multi-valued logic variables in a second relationship are expected to be true; converting data from the process facility into values of multi-valued logic variables; wherein said converting step assigns to the multi-valued logic variables numerical values corresponding to likelihood of truth; calculating the congruence factor for each of the hypothesized plant states in the at least one planned sequence using the multi-valued logic variables; and identifying a current plant state by comparing the congruence factor of each of the hypothesized plant states based and selecting the state having the highest congruency factor.

Claims

1. A method for monitoring operation of a process facility comprising the steps of:

(a) storing at least one planned sequence of hypothesized plant states using multi-valued logic to define calculation of a congruence factor for each of the hypothesized plant states;

wherein said storing in step (a) includes storing relationships between the multi-valued logic variables defined by a maximum value function when any of the multi-valued logic variables in a first relationship may be true and defined by an average function when all of the multi-valued logic variables in a second relationship are expected to be true;

(b) converting data representing the process facility into values of multi-valued logic variables;

wherein said converting in step (b) assigns to the multi-valued logic variables numerical values corresponding to likelihood of truth;

(c) calculating the congruence factor from the multi-valued logic for each of the hypothesized plant states in step (a) in the at least one planned sequence using the multi-valued logic variables; and

(d) identifying a current plant state by comparing the congruence factor of each of the hypothesized plant states to each other and choosing the plant state having the highest congruency factor.

2. A method as recited in claim 1, further comprising the step of (e) displaying information about the current plant state in dependence upon the values of the multi-valued logic variables regardless of whether the information is used in said identifying in step (d).

3. A method as recited in claim 1, wherein said storing in step (a) defines the calculation of the congruence factor for a first set of the hypothesized plant states corresponding to steps in a procedure associated with the planned sequence and a second set of the hypothesized plant states corresponding to predetermined deviations from the first set of the hypothesized plant states.

4. A method as recited in claim 1, wherein said storing in step (a) defines unique calculations to produce a single number as a congruence factor for each of the hypothesized plant states.

5. A method as recited in claim 4,

wherein said identifying in step (d) determines the current plant state as having a maximum congruence factor among the hypothesized plant states, and

wherein said method further comprises the steps of:

(e) temporarily changing the values of the multi-valued logic variables used in calculating the congruence factor of the current plant state, one variable at a time, when the congruence factor of the current plant state is less than a maximum possible congruence factor by more than a predetermined amount;

(f) recalculating the congruence factor for the current plant state for each change in step (e); and

(g) determining how changes in the multi-valued logic variables could improve the congruence factor of the current plant state in dependence upon said recalculating in step (f).

6. A method as recited in claim 1,

wherein said converting in step (b) includes converting the data to the values of the multi-valued logic variables in a range of -1.0 representing certainly false to +1.0 representing certainly true with 0.0 representing total uncertainty, and

wherein the maximum value function used in said storing in step (a) returns an average of the values of the multi-valued logic variables in the first relationship instead of any maximum value within a predetermined amount of 0.0.

7. A method as recited in claim 6, further comprising the step of (e) displaying, when the congruence factor of the current state is less than 1.0 by a predetermined amount, at least one reason from among possible reasons for an imperfect match, the possible reasons including at least one of the multi-valued logic variables, used in the calculation of the congruence factor, having an absolute value not greater than one.

8. A method as recited in claim 1,

wherein said storing in step (a) includes storing a plurality of planned sequences, each of the planned sequences having entrance and exit conditions,

wherein said calculating in step (c) is performed only for the planned sequences having entrance conditions evaluated as true and exit conditions evaluated as false, where true and false are each defined as a range of values in the multi-valued logic, and

wherein said identifying in step (d) is performed to identify the current plant state in each of the planned sequences evaluated by said calculating in step (c).

9. A method as recited in claim 8, further comprising the step of (e) displaying sets of messages, each set corresponding to one of the planned sequences, the messages providing information about the current plant state of corresponding sequence in dependence upon the values of multi-valued logic variables regardless of whether the information is used in said identifying in step (d).

10. A method as recited in claim 1, wherein said storing in step (a) includes defining the calculation of the congruence factor for each of the hypothesized plant states corresponding to steps in a procedure associated with the planned sequence, as including completion of a corresponding step and incompletion of a next step.

11. A method recited in claim 10, wherein said storing step (a) further includes defining the calculation of the congruence factor as also including completion of previous steps.

12. A method as recited in claim 11, wherein said storing in step (a) further includes defining the calculation of the congruence factor as also including physical parameters in expected ranges at the completion of the corresponding step.

13. A method as recited in claim 12, wherein said storing step (a) includes structuring the calculation of each of the hypothesized plant states to give greater weight to the corresponding and next steps than to any of the previous steps and physical parameters.

14. A method as recited in claim 13, wherein said storing in step (a) includes ending the calculation of the congruence factor for each of the hypothesized plant states with the average function.

15. A method as recited in claim 12,

wherein dais identifying in step (d) determines the current plant state as having maximum congruence factor among the hypothesized plant states, and

wherein said method further comprises the steps of:

(e) temporarily changing the values of the multi-valued logic variables used in calculating the congruence factor of the current plant state, one variable at a time, when the congruence factor of the current plant state is less than a maximum possible congruence factor by more than a predetermined amount;

(f) recalculating the congruence factor for the current plant state for each change in step (e);

(g) determining how changes in the multi-valued logic variables could improve the congruence factor of the current plant state in dependence upon said recalculating in step (f); and

(h) displaying information representing the current plant state identified in step (d) and deviations therefrom in dependence upon said determining in step (g) .

16. A method as recited in claim 15, wherein said identifying in step (d) includes checking the hypothesized plant states, having high congruence factors higher than all other hypothesized plant states, for one of the multi-valued logic variables indicating the next step is not completed, when the congruence factors of two of the hypothesized plant states are within a predetermined number.

17. A monitoring system for an electrical power generating plant comprising:

a storage unit to store planned sequences of hypothesized plant states, each plant state defined by a congruence factor calculated using multi-valued logic;

a data acquisition and conversion system to repeatedly obtain plant data and convert the hypothesized plant data representing the electrical power generating station into multi-valued logic variables;

a processor to determine at least one current sequence as likely in progress from among the planned sequences stored in said storage unit, to calculate the congruence factor from the multi-valued logic for each hypothesized plant state in each of the at least one current sequence and to identify a current plant state for each of the at least one current sequence based on a highest congruence factor;

wherein said processor determines the at least one current sequence as having entrance conditions evaluated as true and exit conditions evaluated as false, wherein true and false are each defined as a range of values in the multi-valued logic; and

a display unit to display information about the current plant state in dependence upon the values of the multi-valued logic variables regardless of whether the information is used to identify the current plant state.