中文摘要：

针对仅有位移约束和重量最小的结构拓扑优化问题,基于ICM（独立、连续、映射）方法思路,提出了一种变位移约束限的结构拓扑优化方法.在每一轮子循环迭代求解开始时,为了控制拓扑设计变量的变化量,形成和引进了新的位移约束限.另外,建立了单元删除阈值和几轮迭代循环的单元删除策略.为了确保优化迭代中结构非奇异和方法具有增添单元的功能,在结构孔洞和边界周围引入了一层人工材料单元,并建立了一套有效结构信息到结构最大设计域信息的映射转换方法.结合对偶求解方法,形成了一种新的连续体结构的拓扑优化方法.给出的算例表明该方法没有目标函数的振荡现象,且验证了该方法的正确性和有效性.

英文摘要：

In each sub-loop solving of the ICM （Independent, Continuous and Mapping） method, whether in what quantities do topology design variables change, structural displacements etc. characteristic quantities and their derivatives are approximately obtained by using their values at the beginning of the sub-loop iterations. This measurement may lead to large errors of the mentioned quantity estimations. If there is only a type of constraints （such as displacement constraints） in an optimization model, the errors may be larger. In order to deal with this problem, for the structural topological optimization problem with the objective function being the structural weight and only displacement constraints, this paper proposes a new structural topological optimization method, being based on the ideas of the ICM method （Independent, Continuous and Mapping） and the evolutionary structural optimization method. New displacement constraint limits are formed and introduced to the optimization model at the beginning step of each sub-loop iterations to control variations of topological design variables. Moreover, the element deletion and adding criterion and a set of structural optimization strategy are given. Some elements with artificial material property are inserted around the cavities and boundaries of the structure optimized so that the structure optimized is a non singular structure and the proposed method is of an element restorable function. And a structural characteristics mapping transformation relation between the effective structure and the structural maximum design domain is built. Incorporating the dual programming method, a new continuum structural topological optimization method is proposed. The several examples show that there is not any objective oscillation phenomenon in optimization iterations, and the proposed method is of validity and effectiveness.