基于卷积因子修正的敏度过滤方法  

Sensitivity filter based on different convolution factors methods

作　　者：周利明[1] 傅晓锦[1,2] 

机构地区：[1]上海理工大学机械工程学院,上海200093 [2]上海电机学院机械学院,上海200245

出　　处：《机械设计与制造工程》2015年第4期23-26,共4页Machine Design and Manufacturing Engineering

基　　金：国家自然科学基金资助项目(11202078);上海市教委科研创新重点项目(13ZZ145);上海市自然科学基金资助项目(11ZR1413800)

摘　　要：在变密度法拓扑优化中,敏度过滤是常用的消除数值不稳定的方法之一,但是敏度过滤容易引起模糊边界的产生,为此提出了一种基于卷积因子修正的敏度过滤方法,这种方法通过加强中间单元敏度、减少周边单元敏度的手段,抑制了模糊边界的产生。利用柔度最小化的算例,讨论了这种方法对优化结果的影响,结果证实,此方法可以克服边界模糊和消除棋盘格现象。In the topology optimization of variable density method, sensitivity filter is a common one of the ways to eliminate the numerical instability, But the optimization＇s result of sensitivity filter is easy to cause the fuzzy boundary, In order to prevent this kind of phenomenon of fuzzy boundary, this paper proposes a sensitivity filter based on different convolution factors. This method is to strengthen the middle unit＇s sensitivity value, reduce the peripheral unit＇s sensitivity value, suppress the fuzzy boundary. Using the minimum compliance of the example, it obtains optimization results. The results prove that the method can prevent and alleviate checkerboards, meshdependency and fuzzy boundary.

关 键 词：敏度过滤 插值模型 模糊边界 权重因子 

分 类 号：O343.1[理学—固体力学]