上海交通大学学报

上海交通大学学报（自然版）

• 自动化技术、计算机技术 • 上一篇    下一篇

边界元矩阵稀疏化算法及其应用

崔树标，张云，周华民，李德群   

  1. (华中科技大学 模具技术国家重点实验室， 武汉 430074)
  • 收稿日期:2007-11-10 修回日期:1900-01-01 出版日期:2008-10-28 发布日期:2008-10-28
  • 通讯作者: 崔树标

Research and Application of Sparse Algorithm of
Boundary Element Matrix

CUI Shu-biao, ZHANG Yun, ZHOU Hua-min, LI De-qun   

  1. (State Key Laboratory of Die & Mould Technology,
    Huazhong University of Science & Technology, Wuhan 430074, China)
  • Received:2007-11-10 Revised:1900-01-01 Online:2008-10-28 Published:2008-10-28

摘要： 基于边界元矩阵的空间需求与求解域网格数的平方成正比，提出了边界元矩阵稀疏化方法.首先，根据边界元矩阵的特点定义了合适的稀疏准则，小于该准则的矩阵系数被合并到邻接单元对应的矩阵系数中；然后，将该系数取零，这样可以将一片相互邻接的单元系数合并到其中一个单元，从而达到矩阵稀疏化的目的.仿真结果表明，该方法在保证数值模拟精度的条件下，大幅削减了空间需求.

关键词: 稀疏矩阵, 边界元法

Abstract: A method was brought forward to make the matrix become sparser because the memory space of the boundary element matrix is in direct proportion to the square of mesh number. An appropriate rule of sparsity is defined according to the characteristic of boundary element matrix at first. The matrix coefficients which are less than certain limit value are united into the adjacent elements. Afterward the matrix coefficient was evaluated as zero. And the matrix coefficient of partial adjacent elements can be incorporated into certain element by using this method. Then the boundary element matrix becomes a sparse one. By using this method, the memory space is reduced observably while guaranteeing the numerical simulation accuracy through numerical simulation.

中图分类号: