|Table of Contents|

Neural Network Simulation in Temperature Field Based upon EBE Strategies(PDF)

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

Issue:
2001年05期
Page:
499-503
Research Field:
Publishing date:

Info

Title:
Neural Network Simulation in Temperature Field Based upon EBE Strategies
Author(s):
Yang Zhongchao
School of Mechanical Engineering,Southwest Jiaotong University,Chengdu 610031
Keywords:
f inite element method neural netw ork heat conduct EBE st rategies
PACS:
TB115
DOI:
-
Abstract:
Based upon element-by-element( EBE) iterat ive strateg ies, a feedback mult-i layer neural model is proposed for the heat conduct problem FEM analysis. Linking weig ht coeff-i cients LWC between input-layer and middle-layer is the relat ion matrix A betw een the element sif fness matrix and collect ivity st iffness matrix. LWC betw een the output-lay er and the middlelayer is the t ransposit ive matrix. LWC of the middle-layer super-cells is the element stiffness mat rix about the problem resolved. Numerical simulation show s that the method is consistent w ith FEM. This method present s a new w ay for the parallel analysis by FEM.

References:

1 Hughes T J R. An element- by- element solut ion algorithms for problems of structural and solid mechanics.Comput Meth Appl Mech Eng in, 1983, 36: 241~ 254
2 Takerchi J, Kosugi Y. Neur al networ k representation of finite element method. Neural Network,1994, 7( 2) : 389~ 402
3 孔祥谦. 有限单元法在传热学中的应用. 北京: 科学出版社, 1998
4 周树荃, 梁维泰, 邓绍忠. 有限元结构分析并行计算.北京: 科学出版社, 1997

Memo

Memo:
-
Last Update: 2013-03-25