首页 | 本学科首页   官方微博 | 高级检索  
     检索      

有限元单刚矩阵计算的神经网络法
引用本文:孙道恒,罗萍平.有限元单刚矩阵计算的神经网络法[J].东北大学学报(自然科学版),1997,18(4):431-434.
作者姓名:孙道恒  罗萍平
作者单位:东北大学机械工程学院
摘    要:固体力学有限元计算问题可以采用神经网络进行实时求解,神经网络可用改进的Hopfield网络。求解之前必须确定信号网络的联接权矩阵(相应结构刚度矩阵)。分析了单元刚度矩阵的组成,提出用BP网络来实现单刚矩阵的实时计算方法。分别用传统有限元法和BP网络法求解了一三角形单元的单刚矩阵元素,结果表明BP网络法计算具有实时性,精度满足工程要求。

关 键 词:有限元  刚度矩阵  神经网络  实时计算  计算力学

Neurocomputing Method of Finite Element Stiffness Matrix
Sun daoheng,Luo Pingping,Hu Qiao,Xu Hao.Neurocomputing Method of Finite Element Stiffness Matrix[J].Journal of Northeastern University(Natural Science),1997,18(4):431-434.
Authors:Sun daoheng  Luo Pingping  Hu Qiao  Xu Hao
Abstract:The FE computing problem in solid mechanics can be solved by using the improved Hopfield networks. Before it be solved by using Hopfield net,the weights matrix (i.e.the general stiffness matrix related to an engineering strcture) of the signal net must be defined. So the form of the element stiffness matrix is analyzed,a method which may realize the real time computation of element stiffness matrix based on the BP net is proposed. The element stiffness matrix is computed by using the traditional method and BP net respectively,and it can be seen that the computation is in real time by using BP net,the result is in agreement with the traditional method.
Keywords:finite element  stiffness matrix  neural networks  real  time computation  
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号