基于加速度时程的结构参数直接识别方法及验证

无线智能传感器结合云平台可以实现建筑结构的长期健康监测，模态识别是结构健康监测的重要内容。希尔伯特黄变换(Hilbert-Huang Transform, HHT)因其适用于非线性非平稳信号，且具有完全自适应性等特点，在模态识别领域中被广泛应用。与实验室中进行结构模态参数识别不同的是，长期监测中模态参数识别的算法不能出现主观的参数选择过程，而传统HHT的第一步经验模态分解(Empirical Mode Decomposition, EMD)会产生虚假的固有模态函数(Intrinsic Mode Function, IMF)分量，对虚假分量的识别与剔除往往依赖研究人员的主观判断。该文提出了一种基于深度神经网络(Deep Neural Networks, DNN)与K-L散度(Kullback-Leibler Divergence, K-L Divergence)的新算法，可以自动化识别并剔除EMD产生的虚假分量，从而使得EMD后得到的固有模态函数均为真实分量。

Structural parameters identification methodology with direct use of acceleration time series and its verification

Wireless intelligent sensors combined with cloud storage technology can realize long-term health monitoring for structures. Modal identification is an important part for structural health monitoring. The Hilbert-Huang transform (HHT) is widely used for structural modal identification because it is suitable for nonlinear and non-stationary signals and is self-adaptable. The algorithm of modal identification in the long-term monitoring cannot rely on subjective parameter selection, while the first step of the traditional HHT may produce false intrinsic mode function (IMF) components. The identification and elimination of false IMF components often rely on subjective judgment. In this paper, a new algorithm based on deep neural networks (DNN) and Kullback-Leibler (K-L) divergence is proposed, which can automatically identify and eliminate the false components generated by empirical mode decomposition (EMD).