基于响应协方差小波变换和SVD的结构工作模态参数识别

Operational modal parameter identification based on covariancedriven wavelet transform and singular value decomposition

对系统响应的协方差作小波的时频分解,利用信号互协方差与自协方差的小波变换系数的比值来识别结构的工作模态振型,由矩阵奇异值分解(SVD)从小波变换时频分析结果确定小波脊,通过实际结构多测点数据,利用小波系数比值来反映振型,识别结构各阶工作模态参数(固有频率、阻尼比和振型)。用数值模拟算例和实桥环境振动试验数据对方法进行了验证,并与频域峰值法和时域随机子空间识别方法结果进行了比较,结果表明,该方法可以准确地识别出结构的工作模态参数,特别是阻尼和振型的识别。

The covariance of measured response signals is first converted into the time-scale domain using a continuous wavelet transform. Then, by taking Singular Value Decomposition （SVD） of the covariance matrix, the ridges of the covariance wavelet coefficient are decomposed, which is the extreme value of the eigenvalues. The obtained ridges of the covariance wavelet coefficient magnitudes that represent the main modal features of multi-measurement point signals are used to estimate the operational natural frequencies, damping ratios and modal shapes. A numerically simulated example and a real bridge tested in the field under operational conditions are studied to demonstrate the proposed technique and verify its accuracy with current peak-picking method in frequency-domain and stochastic subspace identification in time-domain. It has been shown that the proposed technique is reliable and efficient to identify the dynamic characteristics of full-size structures under operational conditions, especially the damping ratios and mode shapes.