对偶法求解应力与位移约束不同显式的优化模型

基于响应面法,通过内力将应力约束显式化为正变量的一阶近似,位移约束采用倒变量一阶近似,建立以桁架结构重量最小为目标的截面优化模型,采用对偶规划理论转化为拟无约束模型,采用序列二次规划算法求解。此方法同应力约束零阶近似和一阶近似(使用响应面方法显式化)等方法进行了算例比较,表明本文的优化模型有较好的求解效率。     

基于层次分解方法的桁架结构形状优化

对于桁架结构形状优化,应用层次分解优化方法,将设计变量分成杆件截面积和节点位置两类变量。求解时分为两层,第一层在给定节点位置下对杆件截面进行优化,同时考虑了应力、局部稳定约束和位移约束的重量最轻;第二层假定截面层的有效位移约束作用不变,求解一个使桁架刚度增强的二次规划问题,获得既不违反约束,又使目标函数不上升的新的节点位置,再返回第一层。两层交替进行直至收敛。     

基于响应面方法的结构碰撞优化

应用响应面方法可将结构响应表达为设计变量的显式形式．为拟合响应面，基于中心复合设计和单纯形设计简化出了中心对称和拟单纯形设计，其既可以保证约束近似精度，同时降低计算量．针对常见优化问题，使用响应面方法将其构造为线性或者非线性约束优化模型，并应用Matlab优化工具箱求解．方管碰撞优化数值算例表明文中提出的试验设计方法及优化模型有较好的求解效率和精度．     

桁架结构非概率可靠性形状优化设计

采用区间变量描述不确定参数,提出一种桁架结构非概率可靠性形状优化方法。建立了以截面尺寸和节点坐标为设计变量,以结构重量为目标函数,具有非概率可靠性指标约束的桁架结构形状优化数学模型。采用量纲归一化对截面尺寸和节点坐标进行了变量统一;运用均值点法对功能函数进行泰勒线性近似求解得到相应的非概率可靠性指标,并采用序列二次规划算法对优化模型进行求解。三个算例分析结果表明,算例均能快速稳定地收敛到最优解,结果符合工程结构设计经验,验证了本文所提方法的准确性和有效性。     

近似评价函数确定运动极限及其在形状优化中的应用

针对二维连续体形状优化过程中解析灵敏度求解困难的情况,利用响应面方法将目标函数和约束函数近似显式化,建立序列二次规划模型。为了保证优化模型的可靠性,结合试验设计方法,以较少的结构分析代价构造约束响应面,从而能够快而稳定地收敛,本文重点研究并建立了二次近似评价函数用以计算自适应运动极限的策略。算例说明这种策略是有效而稳定的。     

多约束作用下连续体结构的拓扑优化

基于ICM(独立、连续、映射)方法建立了以结构重量最小为目标,以屈曲临界力、位移及应力三种约束同时作用的连续体拓扑优化模型:采用独立的连续拓扑变量,借助泰勒展式、过滤函数将目标函数作二阶近似展开。借助瑞利商、泰勒展式、过滤函数将屈曲约束化为近似显函数,借助于过滤函数,将位移约束用莫尔定理显式化;将应力这种局部性约束采用全局化策略进行处理,即借助第四强度理论、过滤函数将应力局部性约束转化为应变能约束,大大减少了灵敏度分析的计算量;将优化模型转化为对偶规划,减少了设计变量的数目,并利用序列二次规划求解,缩小了模型的求解规模。数值算例表明,ICM方法在解决屈曲、位移及应力三种约束共同作用的连续体拓扑优化问题上有优势。     

连续体结构屈曲约束的ICM方法拓扑优化

基于ICM(独立、连续、映射)方法解决具有屈曲约束的连续体拓扑优化问题。建立以结构重量为目标,以屈曲临界力为约束的拓扑优化模型;采用独立的连续拓扑变量,借助泰勒展式将目标函数作二阶近似展开;借助瑞利商、泰勒展式、过滤函数将约束化为近似显函数,避免了灵敏度的计算;将优化模型转化为对偶规划,并利用序列二次规划求解,减少了设计变量的数目,缩小了模型的求解规模。给出三个算例,结果表明:该方法可有效地解决屈曲约束的连续体拓扑优化问题,能够得到合理的拓扑结构,并有较高的计算效率。     

基于可靠性的桁架结构拓扑优化设计

建立了以杆截面为设计变量、结构重量极小化为目标、具有位移、应力等性态可靠性约束的桁架结构拓扑优化设计数学模型．通过引入可靠性安全系数，并利用结构力学的三个基本方程，将结构的位移和杆件应力可靠性约束等价显示化为设计变量的线性函数，使原基于可靠性的优化模型转化为常规的序列线性规划问题，利用修正的单纯形法求解．算例表明文中提出的方法既简单又有效．     

考虑性态约束时多工况桁架结构拓扑优化设计

本文提出了一种适用于桁架结构的拓扑优化设计方法。它以杆内力为设计变量,以结构重量为目标函数。该方法的主要特点是:第一,通过引入杆内力为设计变量,既克服了已有方法要求预定位移场这一主要困难,又为在拓扑优化过程中考虑应力、位移等性态约束创造了条件;第二,将多工况的拓扑优化问题描述为一个非光滑的数学规划问题,再通过一个变量代换将其转化为一般的规划为题,进而将原问题的求解又转化为几个线性规划问题的求解;第三,基于结构力学的三个基本方程,将位移与应力约束提成为线性不等式约束,这些约束同重量的目标函数一起构成了拓扑优化设计的线性规划模型。最后,将本方法应用于几个工程算例,得到了满意的数值计算结果。     

应力和位移约束下桁架拓扑优化的有无复合体方法

按照独立连续拓扑变量的思想,将有无复合体模型由构造应力约束问题发展为应力与位移约束问题,用于桁架结构拓扑优化．基于该文提出的位移约束与拓扑变量关系的近似显式,建立了具有应力、位移约束下桁架结构拓扑优化的有无复合体模型,用序列二次规划算法求解．数值结果令人满意．这一工作表明独立连续拓扑变量的提出对于结构拓扑优化的研究是有价值的。     

DESIGN OPTIMIZATION FOR TRUSS STRUCTURES UNDER ELASTO-PLASTIC LOADING CONDITION

In this paper, a method for the design optimization of elasto-plastic truss structures is proposed based on parametric variational principles (PVPs). The optimization aims to find the minimum weight/volume solution under the constraints of allowable node displacements. The design optimization is a formulation of mathematical programming with equilibrium constraints (MPECs). To overcome the numerical difficulties of the complementary constraints in optimization, an iteration process, comprising a quadratic programming (QP) and an updating process, is employed as the optimization method. Furthermore, the elasto-plastic buckling of truss members is considered as a constraint in design optimization. A combinational optimization strategy is proposed for the displacement constraints and the buckling constraint, which comprises the method mentioned above and an optimal criterion. Three numerical examples are presented to show the validity of the methods proposed.     

二次规划的混合能方法及杆系结构弹塑性分析

基于混合能理论构造了二次规划问题求解的新算法，并在杆系弹塑性增量分析中进行了成功的应用。所提出的算法一改传统的对单一规划变量（如位移或应力等）的求解策略，将问题演变成混合变量的形式，在混合空间下对问题进行算法构造，使问题的下一个增量步的求解可以充分利用增量步前的信息，由此极大地提高了算法的求解效率，对杆系弹塑性问题的数值分析充分地证实了这一点。     

Concurrent topology optimization for minimization of total mass considering load-carrying capabilities and thermal insulation simultaneously

The present work introduces a novel concurrent optimization formulation to meet the requirements of lightweight design and various constraints simultaneously. Nodal displacement of macrostructure and effective thermal conductivity of microstructure are regarded as the constraint functions, which means taking into account both the load-carrying capabilities and the thermal insulation properties. The effective properties of porous material derived from numerical homogenization are used for macrostructural analysis. Meanwhile, displacement vectors of macrostructures from original and adjoint load cases are used for sensitivity analysis of the microstructure. Design variables in the form of reciprocal functions of relative densities are introduced and used for linearization of the constraint function. The objective function of total mass is approximately expressed by the second order Taylor series expansion. Then, the proposed concurrent optimization problem is solved using a sequential quadratic programming algorithm, by splitting into a series of sub-problems in the form of the quadratic program. Finally, several numerical examples are presented to validate the effectiveness of the proposed optimization method. The various effects including initial designs, prescribed limits of nodal displacement, and effective thermal conductivity on optimized designs are also investigated. An amount of optimized macrostructures and their corresponding microstructures are achieved.     

Mixed energy method for solution of quadratic programming problems and elastic-plastic analysis of truss structures

A new algorithm for the solution of quadratic programming problems is put forward in terms of the mixed energy theory and is further used for the incremental solution of elastic-plastic truss structures. The method proposed is different from the traditional one, for which the unknown variables are selected just in one class such as displacements or stresses. The present method selects the variables in the mixed form with both displacement and stress. As the method is established in the hybrid space, the information found in the previous incremental step can be used for the solution of the present step, making the algorithm highly efficient in the numerical solution process of quadratic programming problems. The results obained in the examples of the elastic-plastic solution of the truss structures verify what has been predicted in the theoretical analysis. Project supported by the National Natural Science Foundation of China (No. 50178916, No. 19732020 and No. 19872016), the National Key Basic Research Special Foundation (No. G1999032805), the Special Funds for Major State Basic Research Projects and the Foundation for University Key Teachers by the Ministry of Education of China.     

桁架结构形状与尺寸组合优化

提出了一种渐进优化方法 ,通过分离设计变量 ,分别采用渐进节点移动法和满应力法优化桁架结构的形状和尺寸。通过循环迭代 ,耦合形状和尺寸变量间的相互作用关系 ,以期得到结构的最小重量。结构受到应力、局部稳定和节点位移等多种约束。最优解由初始结构设计渐进得到。该方法的有效性通过二个算例得到证实