摘要
借助于“小波”理论,再生核质点方法RKPM(ReproducingKernelParticleMethod)可以将形状函数及求得的结构响应分解为多个尺度。本文对线弹性二维应力集中问题进行了双尺度分解,并由各应力分量计算得到的高梯度点作为误差指示,实现了该方法的h型自适应分析。并且提出了一种新的方法———“四象限法”对高梯度区域进行加密,计算结果表明自适应后的解的精度更高,从而证明了这种自适应无网格方法的有效性。
Borrowing an idea from wavelet theory, the shape function of Reproducing Kernel Particle Method (RKPM) and resultant structural responses are decomposed into different multiple scales. Two-scale decomposition on 2-D linear stress concentration problems is performed in this paper. The obtained highest scale components indicate the high gradient solution and are used as an indicator for h-adaptivity analysis. Furthermore, a new strategy for node refinement based on'four quadrants' criterium is proposed. The numerical results verified the feasibility of this method and also the resultant convergence history demonstrated the convergence of the h-adaptivity.
出处
《强度与环境》
2005年第1期16-20,共5页
Structure & Environment Engineering
基金
国家自然科学基金资助(10202018)
