Multiscale Estimation And Fusion For Structural Dynamics

The wavelet analysis and Kalman filter (KF) were applied in this paper. Based on the multi-scale and model identification theory, the multi-scale data fusion methodology was developed for the multi-sensor, and the multi-rate sensor fusion was also approached.The full state and de-noised structure vibration information was approached, by discrete Kalman filter and data fusion technology. The separately Kalman for several sensors was firstly implemented to obtain the estimation of full-state vibration. Then the distribute data fusion technology was applied to fuse the vibration states estimated by each sensor. The numerical simulations in the paper prove the effectiveness of this estimate fusion strategy.Then, the dynamic system was changed to a multi-scale form using the wavelet multi-scale algorithm, and the Kalman filter was applied in the same scale, and the estimated full-state of the structure was obtained in this scale. After comparison of the rebuilt signals and single scale independent KF data, the de-noise function of wavelet was proved.Finally, the multi-rate multi-scale fusion methodology was developed in this thesis, by using of multi-scale analysis, Kalman filter and distributed data fusion, the numerical simulation approached the data fusion of several multi-rate sensor signals, and the comparison of fused data and original date was also conducted.