基于VMD-SSI的结构模态参数识别

针对结构的弱模态易淹没在噪声中导致模态识别遗漏的问题，该文提出了一种基于变分模态逐步抽取高能量模态的结构弱模态识别方法。该方法采用自回归模型功率谱准确选取高能量模态的初始中心频率；利用变分模态抽取法进行高能量模态分量的分解，接着将去除高能量模态成分的信号代替原始结构响应，重复进行下一阶高能量模态的初始中心频率选取和模态成分分解；对各模态分量进行主成分分析法实现振型识别。进一步，通过地震激励下的十自由度数值算例验证该方法在非平稳激励下弱模态识别的有效性；利用IASC-ASCE健康监测工作组开发的4层框架基准模型试验进行数据分析，验证该方法在实际应用中的有效性。结果表明，所提方法能够用于地震激励下含噪情况的结构弱模态识别。

Application of wavelet transform in structural health monitoring

Because the weak modes of structures are easily drowned in noise and hardly identified, a weak mode identification method is proposed upon repeated variational modal extraction of high energy modes. The autoregressive power spectrum is proposed to select the initial center frequency of the high-energy mode accurately. The high-energy modal component is decomposed by variational modal extraction method. After removing the high-energy component, the residual substitutes the original response for the initial center frequency selection and extraction of the next high-energy component. Mode shape identification is realized upon principal component analysis. A 10-degree-of-freedom numerical example under seismic excitation is presented to verify the effectiveness of the proposed method for weak modal identification under non-stationary excitation. The test data of the benchmark structure proposed by IASC-ASCE health monitoring group is analyzed to verify the effectiveness of the method in practical application. The results show that the proposed method can be used for weak modal identification of structures with noise under earthquake excitations.