2018 Fall Meeting

Presentation information

Oral presentation

III. Fission Energy Engineering » 305-1 Computational Science and Engineering

[3H10-13] Fuel and Reactor Analysis

Fri. Sep 7, 2018 2:45 PM - 3:50 PM Room H (B32 -B Building)

Chair:Takeshi Mitsuyasu(HGNE)

2:45 PM - 3:00 PM

[3H10] Study on Fuel Loading Pattern Intuitive Generation Method by using Deep Learning

(2)Consideration on Application of Convolutional Neural Network

*kazuki ishitani1 (1. Nuclear Engineering and Services Company)

Keywords:fuel loading pattern optimization, deep learning, convolutional neural network, image recognition / image generation

On reload core design, engineer have to find loading pattern(LP) satisfying many of design constraints within a period. In addition, loading fuels should be selected.
Deep Learning technique were proposed to generate LP instantaneously without any try and error to find out worthwhile fuel combination to execute powerful but time consuming optimization tools. Optimized LP's generated by using an ordinary tool were used as training data for convolutional neural network.