节点文献

基于梯度和海森矩阵计算在地震作用下桁架的形状优化

Shape and Cross-section Optimization of Truss Subjected to Earthquake Excitation Based on Calculation of Gradient and Hessian Matrix

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 刘齐茂燕柳斌

【Author】 LIU Qi-mao1,2,YAN Liu-bin2(1.Department of Civil Engineering,Guangxi University of Technology,Liuzhou,Guangxi 545006,China;2.College of Civil and Architecture Engineering,Guangxi University,Nanning 530004,China)

【机构】 广西工学院土木建筑系广西大学土木建筑学院

【摘要】 大型复杂桁架地震响应的形状优化需要大量的计算量,非梯度类算法由于效率低下通常很难成功解决该类问题.本文提出一种在地震作用下以获取质量最小化的二阶优化设计同时满足应力和位移约束的桁架形状优化设计方法.1)在Newmark-β法的基础上导出动力响应及其对设计变量一阶和二阶导数的计算方法;2)通过积分型罚函数将含时间参数的不等式约束问题转变为一系列不含时间参数的无约束问题,并利用动力响应的一阶和二阶导数计算罚函数的梯度和海森矩阵;3)充分利用梯度和海森矩阵的Marquardt方法求解无约束优化问题;演示了一个45杆桁架的形状优化设计.结果表明本文方法是一种桁架在地震作用下有效和高效的形状优化设计方法.

【Abstract】 This paper developed a shape and cross-section optimization method of truss subjected to earthquake excitation for achieving minimum weight design with normal stress and nodal displacement constraints.First,the dynamic responses,their first and second derivatives with respect to design variables are calculated based on Newmark-β method.Second,the inequality constraint problem with time parameter is converted into a sequence of appropriately formed unconstrained problems without time parameter by using the integral penalty function method.The gradient and Hessian matrix of penalty function are calculated by using dynamic response first and second derivatives.Third,Marquardt’s method which makes fully use of gradient and Hessian matrix is employed to solve unconstrained problems. Finally,finding optimum design of a 45-bar truss is demonstrated.The results show that optimization methods presented in this paper are an effective and highly efficient approach for minimum weight design.

【关键词】 桁架形状优化动力响应梯度海森矩阵
【Key words】 trussshape optimizationdynamic responsegradientHessian matrix
【基金】 广西工学院博士基金(03081514)
  • 【文献出处】 北京工业大学学报 ,Journal of Beijing University of Technology , 编辑部邮箱 ,2011年03期
  • 【分类号】TU311
  • 【被引频次】2
  • 【下载频次】124
节点文献中: 

本文链接的文献网络图示:

本文的引文网络