Intrinsic Mode Decomposition Method for Damage and System Identification of In-Situ Civil Infrastructures

传统的数据分析技术已被傅立叶变换所支配。 主要的限制是数据应该是稳定的和足够长的，并且生成数据的系统必须是线性的。 不幸的是，当结构发生损伤，如开裂或屈服时，响应过程是非线性的，因此，基于傅立叶或傅立叶变换的方法，如小波分析方法，不能捕捉到固有的非线性和非平稳特性的物理过程中涉及的振动数据。 NASA古德德太空飞行中心最近开发的一种新技术--经验模态分解（EMD）方法和Hlbert谱分析，已被证明为广泛的工程应用提供了非平稳和非线性数据分析的巨大潜力。 这个项目是一个合作研究之间的加州大学欧文分校和美国宇航局戈达德太空飞行中心。 本文旨在进一步发展最近提出的EMD方法和Hilbert谱分析在民用基础设施损伤检测和系统识别中的应用。 此外，将努力从测量的振动数据中分离出被污染的噪声，并捕获结构响应的内在线性或非线性特性。 还将识别由于结构损伤导致的能量释放产生的特征，并用于表征结构损伤的程度。 提出的独特的方法是基于创新和强大的数据分析技术，捕捉固有的非线性和nonstatinary特性的测量振动数据，系统识别和damagedetection可以。 本研究的成果可望对民用基础设施的系统识别和损伤检测技术的发展起到重要的推动作用。

9807855YangTradtional data Analysis techniques have been dominaed by the Fourier transform. The main restrictions are that the data should be stationary and long enought, and that the system generating the data must be linear. Unfortunately, when the structure is damaged, such as cracking or yielding, the response process in nonlinear, and consequently, Fourier or Fourier-based transform methods such as wavelet analysis approach, is not capable of capturing the intrinsic nonlinear and nonstationary characteristics of the physical process involved in the vibration data. A new technique, the Empirical Mode decompositon (EMD) method and Hlbert spectral analysis, recently developed at NASA Goodard space Flight Center, has been demonstrated to offer a great potential for nonstationary and nonlinear data analyses for wide engineering applications. This project is a collaborative research between the University of California, Irvine, and NASA Goddard space Flight Center. It is aimed at further development of the recently proposed EMD method and hilbert spectral analysis for applications to damage detection and system identification of civil infrastructures. In addition, efforts will be made to separate the polluted noises from measured vibration data and to capture the intrinsic linear or nonlinear characteristics of the structural response. Signatures resulting from the energy release due to structural damage will also be identified and used to characterize the extent of structural damage. The unique approach proposed is based on the innovative and powerful data analysis technique to capture the intrinsic nonlinear and nonstatinary characteristics of the measured vibratin data from which system identification and damagedetection can be made. Teh results of this research project are expected to make a significant advancement of the state-of-the technology in system identification and damage detection of civil infrastructures.