| This thesis presents a wavelet transform-based method of extracting impulse response characteristics from the measured disturbances and response histories of linear structural dynamic systems. Several aspects of wavelets make them a desirable candidate for signal analysis, specifically, orthogonal local basis functions, multiresolution capabilities, and a variety of basis functions for tailoring to the specified problem.;To illuminate its abilities, the proposed wavelet method is closely compared to a more traditional Fourier transform-based procedure. It is shown that for systems subjected to harmonic inputs, the wavelet method is more effective in determining the impulse response function than Fourier procedures. When the system is subjected to more general input signals for which FFT methods are known to perform well, such as random or burst-random signals, the proposed wavelet method performs equally well with a fewer number of ensemble-averages than Fourier-based methods.;Also examined in this thesis are time domain methods of impulse response extraction. The proposed wavelet-based method more closely parallels these procedures due to the nature of the wavelet transform, which retains time information in addition to frequency magnitudes. The major advantage of wavelet-based methods over time domain ones is the ability of wavelets to filter noise present in both the vibration signals and the impulse response functions themselves. No detrimental effects have been shown to result from wavelet denoising, unlike the use of windowing in Fourier procedures, which can induce large amounts of artificial damping in the system.;Numerical examples are used to demonstrate the expressed abilities of the proposed wavelet-based method, mainly through the accuracy of the extracted impulse response functions. System realizations are then performed on the impulse response to find the structural modal frequencies, mode shapes, and damping parameters. Comparison of these values are also used to validate the performance of the wavelet method. Through simulated examples and the analysis of data from an experimental modal test, the wavelet-based structural system identification procedure is shown to be a serious alternative to the more traditional procedures currently used. |
