基于无单元法的发展历史和基本理论,讨论了在无单元法计算中起重要作用的A矩阵的几种取值情况,及其对滑动最小二乘法模拟精度的影响,并修正了滑动最小二乘法计算过程中容易产生数值误差的地方。确定了影响A矩阵的几种极端的布点形式,说明了形函数的值与计算点坐标无关的而只与插值点与计算点的相对坐标相关的性质,并给出了数学理论上或数值上的证明。这对无单元法模拟函数滑动最小二乘法的模拟精度有重要的理论价值和实践意义。
Based on the development of EFG and its basic theory, some circs of matrix in the nun-unit method are discussed and the effect of each circs to simulation of EFG is also analyzed. The case that numerical errors easily come into being in the computable process of the MLSM has been rectified. Some extreme formations of station that affect matrix A are confirmed in this paper, which show the character that the shape function's value does not act with the computing nodes' coordination, but depends on the relative coordination between the interpolation nodes and the computable nodes. These formations are verified by mathematic theory or numerical method, which have important theoretical values and practical significance for the precision of MLSM with the simulation function of EMG.