Modal parameter variation plays an important role in bridge health monitoring. But tracking the long‐term variation of modal parameters is restricted by the modal identifiability. Since some modes cannot be identified due to the limitations of sensor locations or excitation conditions, the identified modes will be missing or misclassification in the tracking process. In this paper, the multistage tracking technique based on subspace correlations is proposed. To avoid missing the identified modes, the reference mode list should be complete to cover all identified modes. Thus, the first stage is to update the reference list adaptively, where the correlations between the observability vectors of identified modes and the subspaces of existing reference modes are taken. The second stage is to cluster each traceable mode and the specified reference mode together without misclassification, where the modal similarity is measured by the similarity of observability vectors (SOV). The proposed algorithm is verified by the vibration data of a numerical model and a highway bridge, respectively. The results show that the proposed method can link with modes in the same order correctly and update the reference mode list adaptively to avoid missing the identified modes.

中文翻译：

---

基于子空间相关性的桥梁模态参数连续跟踪

模态参数变化在桥梁健康监测中起着重要作用。但是，跟踪模态参数的长期变化受到模态可识别性的限制。由于由于传感器位置或激励条件的限制而无法识别某些模式，因此在跟踪过程中，所识别的模式将丢失或分类错误。本文提出了一种基于子空间相关性的多级跟踪技术。为避免遗漏已识别的模式，参考模式列表应完整涵盖所有已识别的模式。因此，第一阶段是自适应地更新参考列表，其中获取已识别模式的可观察性向量与现有参考模式的子空间之间的相关性。第二阶段是将每个可追踪模式和指定的参考模式聚类在一起，而不会发生误分类，其中模式相似性是通过可观察性向量（SOV）的相似性来衡量的。分别通过数值模型和公路桥梁的振动数据验证了该算法的有效性。结果表明，所提出的方法可以正确地以相同顺序与模式链接，并自适应地更新参考模式列表，从而避免丢失识别出的模式。