有限元领域中基于孔特征抑制的模型态生成研究

摘要

The finite element analysis toward complex geometrical model is always a time-consuming task. How to reduce the computation cost with required result precision tums out to be a research focus. The theory of Multi-State Model is introduced to solve this problem, whose key technique is to generate appropriate model state which has critical influence upon computation cost.Since holes in model will contribute largely to huge computation task, based on the technology of feature remove, appropriate simplification strategy toward holes in model is researched, and computation task and result precision of the generated model state are evaluated respectively. A preliminary summary on factors that have relation to result precision and computation task is given in this paper. Multi-Objective problems can be transferred to Single-Objective problems according to classical Multi-Objective resolution, so the most appropriate model state can be gained by using selection algorithm after the establishment of selection criterion.%在有限元领域,对复杂模型进行分析需要耗费巨大的计算资源,如何在保证计算精度的前提下有效提高计算效率一直是一项研究重点.为解决这个问题,引入多态模型理论,通过生成合适的模型态来取得结果精度与计算时间的平衡点.因此,模型态的质量直接关系到计算效率的提升程度.孔特征作为有限元分析模型的常见特征,对计算规模有重要影响.文章基于特征抑制的方式探索模型中孔特征的合适处理策略,并结合孔特征的几何属性佑计不同策略下生成的模型态的计算精度和计算规模.在此基础上找出几何属性之间存在相互制约的关系,根据经典的加权方法将多目标问题转化成单目标问题,建立模型态选择标准并使用遗传算法选择出最优模型态.