Despite numerous studies on vibration reduction techniques for mechanical systems, the expensive computational cost is still a challenging issue for complex structures with a huge number of degrees of freedom. A constructing method of global reduced-order basis (ROB) is proposed in this article by combining proper orthogonal decomposition (POD) with an iterative process. The POD method is adopted to extract the principal component of the snapshot matrix, which collects mode shapes of representative configurations. The objective of the iterative process is to find the global ROB with enough prediction accuracy in the entire design space. Once the global ROB has been constructed, the optimization design for vibration reduction is performed by reduced-order models. Finally, the effectiveness and efficiency of the proposed method are verified through an example of an S-shape stiffened curved shell. The proposed global ROB obtains similar optimal results to the mode-superposition method, with 33.51% computational cost saving.

尽管对机械系统的减振技术进行了大量研究，但对于具有大量自由度的复杂结构而言，昂贵的计算成本仍然是一个具有挑战性的问题。本文将适当的正交分解(POD)与迭代过程相结合，提出了一种全局降阶基(ROB)的构造方法。采用POD方法提取快照矩阵的主成分，收集代表性配置的模态振型。迭代过程的目标是在整个设计空间中找到具有足够预测精度的全局 ROB。一旦构建了全局 ROB，减振优化设计就由降阶模型执行。最后，通过一个 S 形加筋弯曲壳的例子验证了所提出方法的有效性和效率。所提出的全局 ROB 获得了与模式叠加方法相似的最优结果，节省了 33.51% 的计算成本。