Accelerated fixed-point formulation of topology optimization: Application to compliance minimization problems
Accelerated fixed-point formulation of topology optimization: Application to compliance minimization problems
复制标题
DOI：
10.1016/j.mechrescom.2019.103469
复制
发表时间：
2020
影响因子：
2.4
通讯作者：
Weichen Li;Phanish Suryanarayana;G. Paulino
中科院分区：
文献类型：
--
作者：
Weichen Li;Phanish Suryanarayana;G. Paulino
我们提出了一个简单，有效的，可扩展的方法显着加快收敛的拓扑优化模拟。具体而言，治疗的设计过程作为一个固定点迭代，我们建议采用最近开发的加速技术，其中安德森外推定期应用，与简单的加权松弛用于剩余的步骤。通过选定的例子，在遵守最小化，我们表明，所提出的方法是能够加快整体模拟几倍，同时保持解决方案的质量。
We present a simple, effective, and scalable approach for significantly accelerating the convergence in Topology Optimization simulations. Specifically, treating the design process as a fixed-point iteration, we propose employing a recently developed acceleration technique in which Anderson extrapolation is applied periodically, with simple weighted relaxation used for the remaining steps. Through selected examples in compliance minimization, we show that the proposed approach is able to accelerate the overall simulation several fold, while maintaining the quality of the solution.