Cascaded Fuzzy Neural Networks 모델을 이용한 중대사고시 격납용기 압력 예측

Metadata Downloads
Issued Date
cascaded fuzzy neural networks, loss-of-coolant accident, containment pressure
Because of continuously increasing energy demands, many nuclear power plants(NPPs) are in operation globally. The NPPs in long-term operation may be slightly more vulnerable to accidents, such as loss-of-coolant accidents(LOCAs), because the pipes in these plants can be weak. The NPPs automatically operate emergency core cooling systems(ECCSs), such as safety injection system (SIS), when a LOCA occurs. If the SIS is not operating under the severe accident, the core uncovery may occur. When the core uncovery occurs under the accidents such as LOCAs, the containment is difficult to maintain a integrity by the direct heating or hydrogen combustion.
The containment surrounding nuclear steam supply system(NSSS) is one of the facilities that have an important role on nuclear safety in NPPs. The containment is the facility to prevent or minimize leakage of radioactive materials in normal operation or in severe accidents including LOCAs. Therefore, it is important to keep containment integrity. The purpose of this study is to keep the containment integrity by estimating the containment pressure in the event that the SIS is not operating under the severe accident.
The cascaded fuzzy neural networks(CFNN) model is used to predict containment pressure. The CFNN model consists of more than two FNN modules, of which each stage corresponds with a single-stage FNN module. The FNN model is a combination of a fuzzy inference system(FIS) and neuronal training. The CFNN model is developed through the process of repeatedly adding FNN modules. The training of the CFNN model is accomplished by a hybrid method combined with a genetic algorithm and a least squares method.
The CFNN model based on artificial intelligence requires data for its development and verification. Because a variety of real LOCA accident data cannot be obtained from actual NPP accidents, the data used herein were obtained by numerically simulating severe accident scenarios of an optimized power reactor(OPR1000) using MAAP code.
To estimate the containment pressure using the CFNN model, the LOCA break size is used as input data. The LOCA break size cannot be measured, but it was known through previously developed methods. The LOCA break size can be predicted using the trend data for a short time after reactor trip.
It was confirmed that the containment pressure would be exactly predicted by the proposed CFNN model.
Alternative Title
Prediction of the containment pressure under severe accidents using a cascaded fuzzy neural networks model
Alternative Author(s)
Geon Pil Choi
조선대학교 원자력공학과
일반대학원 원자력공학과
Awarded Date
Table Of Contents
목 차

표 목차 ii
그림 목차 iii
Abstract v

제 1 장 서론 1

제 2 장 Cascaded Fuzzy Neural Networks 모델
제 1 절 CFNN 방법론 3
제 2 절 CFNN 모델 최적화 9

제 3 장 CFNN 모델을 사용한 격납용기 압력 예측
제 1 절 CFNN 모델에 사용한 데이터 14
제 2 절 실험결과 16

제 4 장 결론 40

【참고문헌】 41
조선대학교 대학원
최건필. (2016). Cascaded Fuzzy Neural Networks 모델을 이용한 중대사고시 격납용기 압력 예측.
Appears in Collections:
General Graduate School > 3. Theses(Master)
Authorize & License
  • AuthorizeOpen
  • Embargo2017-02-21
Files in This Item:

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.