The COC algorithm, part I: Cross-section optimization or sizing
The COC algorithm, part I: Cross-section optimization or sizing
复制标题
DOI：
10.1016/0045-7825(91)90045-8
复制
发表时间：
1991-08
影响因子：
7.2
通讯作者：
G. Rozvany;M. Zhou
中科院分区：
文献类型：
--
作者：
G. Rozvany;M. Zhou
在迭代连续最优性准则（COC）方法中，最优性的必要条件首先使用变分方法解析生成，然后重新解释为虚构或“伴随”结构。最后，对真实的结构和伴随结构进行离散化，并采用有限元（FE）方法迭代求解。它示出的“FE模拟器”的特殊类的问题，COC优化器可以潜在地处理数以百万计的变量的横截面优化（或尺寸），从而不仅消除，而且扭转了目前的分析能力和优化能力之间的差异。
In iterative continuum-based optimality criteria (COC) methods, necessary conditions of optimality are first generated analytically using variational methods and then reinterpreted in terms of a fictitious or ‘adjoint’ structure. Finally, the real and adjoint structures are discretized and the problem is solved by an iterative procedure using finite element (FE) methods. It is shown by means of ‘FE simulators’ for special classes of problems that a COC optimizer could potentially handle millions of variables in cross-section optimization (or sizing), thereby not only eliminating but also reversing the present discrepancy between analysis capability and optimization capability.