| A new technique for operational modal analysis in frequency domain is proposed in this thesis, with a focus on the complex large-scale structures such as high buildings and long-span bridges, which are difficult to be excited artificially. The main contents are as the following: 1) Traditional random decrement technique (RDT) under four triggering conditions (level crossing triggering condition, local extremum triggering condition, zero crossing triggering condition and positive point triggering condition) is carefully studied. The corresponding random decrement functions (RDF) are then compared with the correlation functions. 2) A new kind RDT, vector random decrement technique (VRD) is studied and realized, which has a advantage in the computational efficiency. 3) The time-domain RDFs are transformed to frequency-domain half power spectral density functions (HPSD) by the Fourier Transformation, and then the modal parameters including modal frequencies, damping ratios, and mode shapes are identified from the HPSDs, with the classical identification algorithm named as Complex Mode Indicator Functions (CMIF). 4) All the algorithms are realized in Matlab, and a simulation example of three-dimensional space structure is utilized to validate the algorithms. Finally, the algorithms are applied to analyze the real measurement data from Chaobai river bridge in Beijing.The method presented in this thesis is more efficient than the traditional operational modal identification methods. Moreover, the realization has much more flexibility thanks to the flexible selection of different triggering conditions and triggering levels. Additionally, besides CMIF, most classical experimental modal analysis methods based on frequency response function in frequency domain can be employed to extract modal parameters from the HPSD, due to the similarity between each other. This method bridges the gap between operational modal analysis and experimental modal analysis.
|
PDF Full Text Request