Loading...

Design of Transient Identification Tool using Deep Learning in PWR Nuclear Power Plant (case study: Bushehr Nuclear Power Plant)

Ramezani, Iman | 2023

26 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 56760 (46)
  4. University: Sharif University of Technology
  5. Department: Energy Engineering
  6. Advisor(s): Vosoughi, Naser; Ghofrani, Mohammad Bagher
  7. Abstract:
  8. In a nuclear power plant, transients are initiated for various reasons such as equipment failure or external disturbances. A transient should be identified as quickly as possible so that countermeasures can be taken to reduce its negative consequences. Due to the large number of parameters that can be monitored in nuclear power plants, the time limit for interpreting the information, and the stressful conditions of the incident, it will be very difficult to detect transients by the plant operators. Therefore, the development of operator support tools to identify transients is of great importance in the safe operation of nuclear power plants. Various studies have shown that data-driven methods and especially artificial neural networks have been effective in this field. Considering the proper performance of deep networks in solving problems with high complexity and large dimensions of input data, in this thesis the efficiency of modern deep learning methods in diagnosing transients of pressurized water reactor (PWR) nuclear power plants has been studied. For this purpose, at first, the most effective power plant parameters that should be used as network input was determined by feature selection methods. Then, by combining convolutional deep networks and long short-term memory, a hybrid CNN-LSTM deep network was developed with a semi-online approach and used to detect power plant transients. The efficiency of the proposed deep network was evaluated in identifying the transient type, determining the location of the initiating event, and estimating its severity, as well as predicting the future trend of the important parameters of the power plant. Bushehr Nuclear Power Plant is the case study of this research and the data set of transients used was extracted from the full-scope simulator of Bushehr Nuclear Power Plant. The results showed that the proposed method performed well in terms of accuracy, identification time, and computational cost, which makes it possible to use it in an online transient identification tool with practical application. Finally, to aggregate the conducted studies in the form of an operator support tool, the developed sections in this thesis were introduced as a structure consisting of different modules, and the conceptual design of an operator support tool for identifying transients of the nuclear power plant was presented. The use of this tool will lead to the improvement of the response time of the operators to accidents and reduce the possibility of human error in response to the transients, which will increase the safety in the operation of the nuclear power plant and avoid economic losses
  9. Keywords:
  10. Transient Identification ; Deep Learning ; Long Short Term Memory (LSTM) ; Convolutional Neural Network ; Bushehr Nuclear Power Plant ; Pressurized Water Reactor (PWR,WWER) ; CNN-BiLSTM Deep Neural Networks

 Digital Object List

 Bookmark

No TOC