【摘要】 运营模态分析在状态评估、损伤检测和模型更新等结构健康监测领域中发挥着重要作用。然而，受网络延迟、传感器误差和设备故障等多种因素的影响，多通道的结构振动监测信号往往存在轻微或显著的不同步，从而产生相位偏差，导致模态分析过程中出现难以预测的不确定性。为此，提出了一种改进的频域分解法，用于检测和同步具有任意时延的多通道振动响应数据并识别结构模态参数。该方法基于频域分解法互功率谱密度矩阵，提取各通道响应信号在频域中的相位变化特征，并将时延转换为特定带宽内的相位周期数之差，获得时延的解析值。通过融合多个模态间的相位信息，有效降低系统误差及噪声的影响，从而显著提升时延估计的准确性。最后，分别利用一个6自由度数值模型以及苏通大桥和上海中心大厦的实测加速度响应数据验证了该方法的有效性。结果表明，该方法执行快、精度高，既可用于非同步数据下的模态参数识别，又可用于振动信号时延的检测与估计。更多还原

【Abstract】 Operational modal analysis plays an important role in structural health monitoring fields of state assessment, damage detection and model updates. However, due to various factors of network delays, sensor errors and equipment failures, multi-channel structural vibration monitoring signals often have slight or significant asynchrony to cause phase deviations and unpredictable uncertainties in modal analysis process. Here, an improved frequency domain decomposition method was proposed for detecting and synchronizing multi-channel vibration response data with arbitrary time lags, and identifying structural modal parameters. This method could be based on the cross power spectral density matrix of frequency domain decomposition method to extract phase change characteristics in frequency domain of various channels’ response signals, convert time lag into phase period differences within a specific bandwidth, and obtain analytical value of time lag. By integrating phase information among multiple modes, effects of system errors and noise were effectively reduced to significantly improve the correctness of time-lag estimation. Finally, a 6-DOF numerical model and the measured acceleration response data of Sutong Bridge and Shanghai Tower were used to verify the effectiveness of the proposed method. The results showed that the proposed method executes quickly, has high accuracy and can be used for modal parameter identification under asynchronous data, as well as for detection and estimation of time lags in vibration signals.更多还原