Method for finding cause of abnormal event in support of plant operation

• US 5,799,294 A
• Filed: 02/23/1996
• Issued: 08/25/1998
• Est. Priority Date: 02/28/1995
• Status: Expired due to Fees

• Current Assignee: Mitsubishi Jukogyo Kabushiki Kaisha
• Original Assignee: Mitsubishi Jukogyo Kabushiki Kaisha
• Inventors: Tanaka, Satoshi
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): Rhodes, Jason W.
• Application Number:
  US08/606,088
• Time in Patent Office: 914 Days
• Field of Search: 395/51, 395/52, 395/54, 395/60, 395/63, 395/61
• US Class Current: 706/46
• CPC Class Codes: G05B 23/0278 Qualitative, e.g. if-then rules; Fuzzy logic; Lookup tables; Symptomatic search; FMEA

Abstract

The accuracy of finding a cause from factors cited by means of a MYCIN method by narrowing down such factors is improved. With respect to deriving events which would not possibly occur if a particular factor 10C among a plurality of factors is a cause of abnormal event 100C, negative deriving events 12C and 1MC negating these deriving events of the other factors which would not occur with factor 10C being the cause are added to a fault tree chart as deriving events of factor 10C.

Claims

1. A method performed in a plant operation supporting device having a computer for finding a cause of an abnormal event in support of a plant operation comprising the steps of:

using the plant operation supporting device, detecting an abnormal event in the plant operation;

using a relationship between the abnormal event and a plurality of factors causing the abnormal event, a fault tree chart showing relationships among the factors and deriving events thereof, and an event evaluation value indicating a possibility of occurrence of the deriving events;

with respect to a first deriving event which is associated with a first factor among said plurality of factors and which would not occur if a second factor among said plurality of factors which is different from the first factor is the cause of the abnormal event, providing a second deriving event which is obtained by taking the negative form of said first deriving event; and

adding said second deriving event to said fault tree chart;

wherein when an event evaluation value for each deriving event is V_ij, a certainty CF_i of each factor is calculated using the following equation CF_i =R_i ·

V_i1^Ri1/R ·

V_i2^Ri2/R. . . V_iM^RiM/R wherein R=R_i1 +R_i2 +. . . +R_iM and R_in is a set value indicating the strength of a relationship between each deriving event and each factor; and

wherein a factor having a largest value among values of certainties CF_i is identified as the cause of the abnormal event.

2. A method according to claim 1, wherein said adding step comprises the step of:

adding said second deriving event to said fault tree chart as a deriving event of said second factor.

3. A method according to claim 2, wherein event evaluation values are assigned corresponding to the occurrence and non-occurrence of the respective deriving events and said identified factors.

4. A method according to claim 3, further comprising a determining step which calculates a certainty value for each identified factor, and wherein existence of deriving events which would not occur if an identified factor is the cause of the abnormality reduces the certainty value for that factor.

5. A method according to claim 4, wherein the factor having the highest certainty value is identified as the cause of the abnormal event.

6. A method performed in a slant operation supporting device having a computer for determining a cause of an abnormal event in a plant operation, comprising the steps of:

monitoring predetermined parameters for abnormal events in the plant operation;

detecting an abnormal event in the plant operation;

identifying a plurality of factors possibly causally related to said abnormal event in the plant operation defined by a fault tree;

comparing said possible factors with occurrence and non-occurrence of a plurality of deriving events;

assigning event evaluation values corresponding to the occurrence and non-occurrence of the respective deriving events and said identified factors;

negating deriving events for identified factors associated with non-occurrence of said deriving events by adding deriving events in negative form to said fault tree; and

determining a probability of fault of each of said factors corresponding to the occurrence and non-occurrence of said deriving events;

wherein when an event evaluation value for each deriving event is V_ij, a certainty CF_i of each factor is calculated using the following equation CF_i =R_i ·

V_i1^Ri1/R ·

V_i2^Ri2/R. . . V_iM^RiM/R wherein R=R_i1 +R_i2 +. . . +R_iM and R_in is a set value indicating the strength of a relationship between each deriving event and each factor; and

wherein a factor having a largest value among values of certainties CF_i is identified as the cause of the abnormal event.

7. A method according to claim 6, wherein said fault tree chart defines relationships among the factors and respective deriving events thereof.

8. A method according to claim 7, wherein said determining step calculates a certainty value for each identified factor, and wherein existence of deriving events which would not occur if an identified factor is the cause of the abnormality reduces the certainty value for that factor.