2017 Fall Meeting

Presentation information

Oral presentation

III. Fission Energy Engineering » 307-1 Computational Science and Engineering

[1G09-12] Seismic and Fuel Analysis

Wed. Sep 13, 2017 2:45 PM - 3:50 PM Room G (C213 - C Block)

Chair:Akemi Nishida (JAEA)

3:15 PM - 3:30 PM

[1G11] Study on Fuel Loading Pattern Optimization using Deep Learning and Reinforcement Learning

(2)Study on Prediction of Depletion Characteristics of Reactor Core

*Masahiro Tatsumi1 (1. NEL)

Keywords:Deep Learninng, Nuclear Characteristics of Reactor Core, Depletion Calculation, Reinforcement Learning, Loading Pattern Optimization

We constructed a deep neural network with infinite multiplication factor vector as input and tried to evaluate the depletion characteristics of the core. It is confirmed that it is possible to predict the depletion characteristics of the maximum power of fuel assembly and the critical boron concentration at high speed and made uncertainty assessment. In some cases, the prediction error is large, so further investigation is necessary.