Plant malfunction diagnostic method

• US 5,023,045 A
• Filed: 02/07/1990
• Issued: 06/11/1991
• Est. Priority Date: 02/07/1989
• Status: Expired due to Term

• Current Assignee: Japan Nuclear Cycle Development Institute
• Original Assignee: Doryokuro Kakunenryo Kaihatsu Jigyodan
• Inventors: Watanabe, Kenshiu, Tamayama, Kiyoshi
• Primary Examiner(s): Walsh, Donald P.
• Application Number:
  US07/476,922
• Time in Patent Office: 489 Days
• Field of Search: 376/215, 376/216, 376/217, 364/148, 364/274.9, 364/275.2, 364/495, 364/513, 364/972.4
• US Class Current: 376/215
• CPC Class Codes: G21D 3/04 Safety arrangements emergency protection of reactor G21C9/00
• CPC Class Codes: Y02E 30/00 Energy generation of nuclear origin
• CPC Class Codes: Y02E 30/30 Nuclear fission reactors
• CPC Class Codes: Y10S 706/914 Process plant

Abstract

A plant malfunction diagnostic method is characterized by determining by simulation a change in a plant state variable, forming a pattern among plant state variables obtained by autoregressive analysis of the change in plant state variable, inserting the formed pattern among the plant state variables in a neural network, performing learning until a preset precision is obtained, and identifying the cause of the malfunction by inserting, in the neural network, a pattern which indicates the pattern among plant state variables formed by data gathered from the plant. This makes possible early identification of the cause of a malfunction. Plant rate of operation and safety are improved by allowing the operator to perform the appropriate recovery operation with a sufficient time margin.

Claims

1. A plant malfunction diagnostic method comprising:

a step of determining by simulation a change in a plant state variable at the time of a malfunction and forming a pattern among plant state variables obtained by autoregressive analysis of the change in plant state variable;

a step of inserting the formed pattern among the plant state variables in a neural network comprising an input layer, a middle layer and an output layer, and performing learning within a preset precision to decide a model of the neural network as well as connection weight between processing elements (neurons) in the input layer and middle layer and between the middle layer and the output layer; and

a step of identifying the cause of the malfunction by inputting a pattern, which indicates the pattern among plant state variables formed from the plant state variable, after detection of an actual plant malfunction.

2. The method according to claim 1, wherein the pattern among said plant state variable is a coherency function indicating a correlation of the frequency spectrum of each plant state variable.