Wavelet-based identification of time-varying shear-beam buildings using incomplete and noisy measurement data

Quite a lot of structural identification techniques assume that the complete state of the structure can be measured which unfortunately cannot be easily met in practice. In addition, measurement data are normally contaminated with noise which unavoidably affects the performance of these techniques. In this paper, two waveletbased identification techniques are developed to identify time-varying parameters in a shear-beam building using incomplete and noisy measurement data. By approximating the unknown time-varying parameters using wavelet multiresolution expressions, a parametric identification problem is transformed into a nonparametric coefficient optimization. As the wavelet multi-resolution expressions can accurately represent arbitrary functions, these wavelet-based techniques can be used to identify various forms of timevarying behavior that can occur to the building’s parameters. Depending on the availability of measurement data, two wavelet-based identification techniques, story-by-story and sub-structural, are proposed. To minimize the effect of measurement noise, a wavelet decomposition level selection scheme and a multi-rate Kalman data fusion technique are adopted. Numerical studies are conducted to illustrate the performance of the proposed wavelet-based timevarying structural identification techniques. Results show that the techniques can track the time-varying properties of shear-beam buildings quite accurately under incomplete and noisy measurement condition