Modal Identification of Various Types of Structures Using Frequency Domain Decomposition

• Main Authors: You-Cheng Luo, 羅友晟
• Other Authors: Gwolong Lai
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/x848ru

碩士 === 國立雲林科技大學 === 營建工程系 === 104 === With the advance of monitoring technology and system identification of structures, structural health monitoring has become an important field in both academic and industrial circles. In recent years, system identification of large structures is often accomplished based on ambient vibration measurements. However, only the output response of the structure can be obtained from ambient vibration measurements. Without any information about the input forces, output-only modal analysis, which is also called operational modal analysis (OMA), is needed to identify the modal parameters of structures. Several OMA methods have been developed and applied to various types of structures, for example, the peak picking method, the stochastic subspace identification techniques (SSI) and the frequency domain decomposition method (FDD), etc. In this study, the frequency domain decomposition method is examined and exploited in depth. After converting the decomposed SDOF power spectral density function (PSD) to the time domain, the modal properties are found using an improved identification method. Furthermore, in order to evaluate the optimal parameters in the FDD method for different types of structures, numerical simulation models are analyzed first and the identified modal properties are compared with the exact values. Then the modal analysis of real building and bridge structures are carried out using FDD based on ambient vibration measurements, and the results are compared with those using SSI. This study shows that all of the identified modal parameters by FDD are close to the exact values or the results using SSI, which means that FDD is an effective method of OMA. In addition, the computation of FDD is relatively simple and fast compared to other OMA methods. The modes with close frequencies can be separated and readily identified by FDD.