多塔斜拉桥加劲索涡激振动实测与时域解析模态分解

为研究多塔斜拉桥中塔加劲索涡激振动时域和频域特性，对多根加劲索开展了振动加速度测量和风速、风向观测，研究了加劲索振幅与风速和风向的关系，分析了加速度时程的时域和频域特征。采用解析模态分解法对加劲索涡激振动加速度时程进行了分解，分析了所得分量的时域和频谱特征。研究发现，在无雨和较低风速条件下，同侧并列加劲索仅迎风侧发生明显涡激振动，其峰值振动是以频率为6.25 Hz的第28阶模态主要参与为特征，为高阶多模态涡激振动，明显发振风速约为4~5 m·s^-1，风向接近垂直桥轴线，其面内振动明显大于面外。1^#加劲索面内涡激振动时程分解得到的3个相邻高阶频率时程分量显示，第28阶模态振动加速度随时间的变化，主导了加劲索振动加速度幅值的增大和减小。同时认为，解析模态分解法不仅能较好地分离含有多个密集频率分量的时域信号，且分解得到的分量不改变原信号分量的频率特征，分解分量再合成的信号与原信号时频特征完全一致。因而可采用解析模态分解法分解具有多个密集频率分量的柔性结构响应，能有助于工程结构风致响应的模态参数识别。

Field Observation of Vortex-induced Vibration of Stiffening Cables in a Multi-tower Cable-stayed Bridge with Application of Analytical Mode Decomposition

To investigate the vibration type and time- and frequency-domain behavior of vortex-induced vibration of stiffening cables in a multi-tower cable-stayed bridge, field measurements of vibration acceleration of stiffening cables, wind speed, and yaw angle at the bridge site were taken. The relationship between cable acceleration, wind speed, and wind yaw angle was presented, and the time- and frequency-domain behavior of acceleration records was investigated. Analytical mode decomposition (AMD) was subsequently employed to decompose the measured records and to further investigate the time- and frequency-domain properties of closely spaced frequency components. Under weather conditions of no rain and low wind speed, significant vortex-induced vibration (VIV) was observed only on the windward cable of the closely spaced twin stiffening cables, and the peak vibration is characterized by the dominating participation of the 28th mode with a frequency of 6.25 Hz. The VIV is also characterized as a joint participation of multiple high-order modes, with typically wind speed ranges from 4 to 5 m·s^-1 and wind yaw angle approximately normal to the bridge axis, with the in-plane peak acceleration significantly larger than that of the out-of-plane component. The three decomposed components, corresponding to three sequent high-order modals in the in-plane VIV signal of stiffening cable No. 1, indicate that the components corresponding to the 28th mode dominate the growth and decay of vibration. This indicates that not only can AMD successfully decompose the time-domain signal with closely spaced frequency but also the decomposed components will not change the frequency properties of the original signal. Meanwhile, the decomposed-synthesized signal is in perfect agreement with the original signal. Based on this study, it is concluded that AMD is capable of decomposing the flexible structure response with closely spaced frequency components and hence can facilitate identification of modal parameters of engineering structures.