基于模型缩减方法的抗震结构动力学拓扑优化

DYNAMIC TOPOLOGY OPTIMIZATION OF SEISMIC STRUCTURES BASED ON MODEL REDUCTION METHOD

  • 摘要: 该文针对地震加速度激励下连续体结构抗震设计问题,提出了一种基于模型缩减方法的高效拓扑优化方法,建立了材料体积约束下最大动柔度最小化的拓扑优化模型。为了确保目标函数的可微分性质,采用广义均值函数等效表示时域内的最大动柔度。基于改进的参数化水平集带方法实施拓扑优化,准许基于梯度的优化算法对优化问题进行求解。结合直接法和伴随向量法详细推导了最大动柔度和体积约束关于设计变量的导数,并基于移动渐近线方法(method of moving asymptotes, MMA)更新设计变量。为了提高计算效率,将基于准静态里兹向量(Quasi-Static Ritz Vector, QSRV)模型缩减技术引入到Newmark方法中,实现瞬态动力学方程及灵敏度分析中伴随方程的高效求解。最后,通过人工地震作用下的三个典型算例来说明该文方法的有效性。同时,比较了基于QSRV缩减方法和Newmark全分析方法的优化结果。结果表明基于QSRV缩减技术的拓扑优化方法能大幅度缩短瞬态动力学优化迭代的计算时间,提高计算效率。

     

    Abstract: An efficient model reduction-based topology optimization method is developed for solving the design problem of continuous structures under seismic acceleration. A topology optimization model of minimizing the maximum dynamic compliance under material volume constraint is established. To ensure the differentiability of the objective function, the generalized mean function is used to represent the maximum dynamic compliance in the time domain. Topology optimization is carried out using the enhanced parametric level set band method, which allows gradient-based optimization algorithms to solve optimization problems. The derivatives of maximum dynamic compliance and volume constraint with respect to design variables are detailed by combining the direct method and adjoint vector method. Subsequently, the method of moving asymptotes (MMA) is employed to update design variables. To enhance the efficiency of optimization calculation, a model reduction technique based on Quasi-Static Ritz Vector (QSRV) is incorporated into the Newmark method to achieve efficient solution of transient dynamics equations and adjoint equations in sensitivity analysis. Finally, three typical examples under artificial earthquake are given to illustrate the effectiveness of the proposed method. At the same time, the optimization results based on the QSRV reduction method and the Newmark total analysis method are compared. The results indicate that the topology optimization method based on QSRV reduction technology can significantly reduce the calculation time of transient dynamic optimization iteration and improve the calculation efficiency.

     

/

返回文章
返回