Expert system for surveillance, diagnosis and prognosis of plant operation

• US 5,009,833 A
• Filed: 01/11/1989
• Issued: 04/23/1991
• Est. Priority Date: 01/11/1989
• Status: Expired due to Fees

• Current Assignee: Westinghouse Electric Company LLC (Brookfield Asset Management Incorporated)
• Original Assignee: Westinghouse Electric Company LLC (Brookfield Asset Management Incorporated)
• Inventors: Takeuchi, Kenji, Meyer, Philip E., Gagnon, Andre F., Cheung, Augustine C.
• Primary Examiner(s): Walsh, Donald P.
• Application Number:
  US07/295,698
• Time in Patent Office: 832 Days
• Field of Search: 376/215, 376/216, 376/217, 364/138 X, 364/492, 364/188, 364/189, 364/431.01, 364/513, 364/556, 364/578, 364/274.5, 364/917.94
• US Class Current: 376/217
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models
• CPC Class Codes: Y04S 10/50 Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
• CPC Class Codes: Y10S 706/914 Process plant

Abstract

An expert system is used to perform surveillance, diagnosis and prognosis of a plant such as a nuclear reactor used in electrical power generation. The expert system uses one of current actual plant data, previously stored actual plant data or simulated plant data as plant condition data to perform an evaluation of operating conditions in the plant. When current actual plant data is used, a data acquisition system supplies the plant condition data. The expert system accesses the plant condition data, regardless of its source, periodically as a surveillance frame is triggered by the rules in the root frame controlling the expert system. Other frames in the expert system perform diagnosis and prognosis of the plant condition.

Claims

1. A method for analyzing operating conditions in a plant, comprising the steps of:

(a) obtaining plant condition data indicative of the operating conditions in the plant;

(b) evaluating the plant condition data using an expert system rule base to determine probabilities of existence of abnormal circumstances;

(c) predicting effects of the abnormal circumstances in dependence upon the plant condition

2. A method as recited in claim 1, wherein step (a) comprises at least one of:

(a1) monitoring operational units in the plant to obtain current actual plant data as the plant condition data concurrently with execution of step (b);

(a2) inputting previously recorded actual plant data as the plant condition data; and

(a3) receiving simulated plant data as the plant condition data.

3. A method as recited in claim 2, further comprising the steps of:

(d) providing, as alternatives for said obtaining, said monitoring of step (a1), said inputting of step (a2) and said receiving of step (a3); and

(e) selecting one of said monitoring of step (a1), said inputting of step (a2) and said receiving of step (a3) to obtain the plant condition data.

4. A method as recited in claim 3, wherein said evaluating in step (b) includes the step of periodically updating the plant condition data using the current actual plant data obtained by said monitoring in step (a).

5. A method as recited in claim 2, wherein step (b) comprises:

(b1) chaining rules backward to determine possible causes of the operating conditions; and

(b2) determining probable occurrence of events using forward chaining logic.

6. A method as recited in claim 2, wherein step (b) comprises:

(b1) determining whether the plant condition data indicates that an operation effecting event has occurred;

(b2) evaluating the plant condition data for the abnormal circumstances unlikely to cause the operation effecting event within a predetermined amount of time; and

(b3) evaluating the plant condition data for the abnormal circumstances capable of causing the operation effecting event after the operation effecting event is detected in step (b1).

7. A method as recited in claim 6, wherein the plant is a nuclear reactor,

wherein the operation effecting event includes a reactor scram and accidents occurring without causing a reactor scram, and

wherein step (b) further includes the step of (b4) determining existence of other abnormal circumstances undetected in step (b1).

8. A method of analyzing operating conditions in a plant, comprising the steps of:

(a) providing alternative methods for obtaining plant condition data indicative of the operating conditions in the plant, the alternative methods including

(a1) monitoring operational units in the plant to obtain current actual plant data as the plant condition data;

(a2) inputting previously recorded actual plant data as the plant condition data; and

(a3) receiving simulated plant data as the plant condition data;

(b) selecting one of said monitoring of step (a1), said inputting of step (a2) and said receiving of step (a3) to obtain the plant condition data; and

(c) evaluating the plant condition data, concurrently with said monitoring in step (a1), using an expert system rule base to determine probabilities of existence of abnormal circumstances.

9. A method as recited in claim 8, wherein step (c) comprises:

(c1) chaining rules backward to determine possible causes of the operating conditions; and

(c2) determining probable occurrences of events using forward chaining logic.

10. A method as recited in claim 8, wherein step (c) comprises:

(c1) determining whether the plant condition data indicates that an operation effecting event has occurred;

(c2) evaluating the plant condition data for the abnormal circumstances unlikely to cause the operation effecting event within a predetermined amount of time; and

(c3) evaluating the plant condition data for the abnormal circumstances capable of causing the operation effecting event after the operation effecting event is detected in step (b1).

11. A method as recited in claim 10, wherein the plant is a nuclear reactor,

wherein the operation effecting event includes a reactor scram and accidents occurring without causing a reactor scram, and

wherein step (b) further includes the step of (b4) determining existence of other abnormal circumstances undetected in step (b1).

12. An apparatus for analyzing operating conditions in a plant, comprising:

data input means for obtaining plant condition data indicative of the operating conditions in the plant by continuously monitoring operational units in the plant to obtain current actual plant data as the plant condition data;

storage means for storing an expert system rule base; and

evaluation means for evaluating, concurrently with the monitoring performed by said data input means, the plant condition data using the expert system rule base to determine probabilities of existence of abnormal conditions.

13. An apparatus as recited in claim 12, wherein said evaluation means is controlled by the expert system rule base to periodically update the plant condition data using the current actual plant data obtained by said monitoring means.

14. An apparatus as recited in claim 13,

further comprising:

prior data means for storing previously recorded actual plant data; and

simulation means for supplying simulated plant data, and

wherein said data input means comprises means for obtaining the plant condition data from one of the current actual plant data, the previously recorded actual plant data and the simulated plant data under control of said evaluation means.