합성곱 신경망을 이용한 건물 온도 분포 및 열 손실 예측Prediction of temperature distribution and heat loss of building via convolutional neural network
- Other Titles
- Prediction of temperature distribution and heat loss of building via convolutional neural network
- Authors
- 김덕중; 김학성
- Issue Date
- Jun-2021
- Publisher
- 대한기계학회
- Citation
- 대한기계학회 재료 및 파괴부문 2021년도 춘계학술대회 논문집, pp.88 - 88
- Indexed
- OTHER
- Journal Title
- 대한기계학회 재료 및 파괴부문 2021년도 춘계학술대회 논문집
- Start Page
- 88
- End Page
- 88
- URI
- https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/191341
- Abstract
- In this study, deep learning method was developed to predict temperature distribution and thermal loss of building considering thermal bridge. Previous studies developed a various kind of analytical method to evaluate energy performance of buildings. However, discontinuity of structures and thermal properties difference of building materials cause huge amount of thermal loss due to thermal bridge, and it was difficult to be predicted. Therefore, finite element method has been developed by previous researchers [1]. However, it hasstill limitation that simulation model should be made in every case considering its complex structures and thermal properties, which have a wide range. In addition, expert knowledge and technique for simulation is required for accurate prediction with high reliability. In this study, FEM database with a different geometries and material properties was constructed and
trained by deep neural network based on convolutional neural network. With the pre-trained deep neural network, we can obtain the temperature distribution and total thermal loss considering thermal bridge in a few seconds by only entering blueprint and thermal properties. It is expected that the developed model can be used for various applications to train finite element analysis.
- Files in This Item
-
Go to Link
- Appears in
Collections - 서울 공과대학 > 서울 기계공학부 > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.