Structural Health Monitoring Based On GPS Technology Under Ambient Excitation

During the service time, large civil engineering infrastructures inevitably suffer from environmental corrosion, material aging, fatigue, impact and the coupling effects with long-time loads and extreme loads. The induced damage accumulations and performance degenerations would seriously reduce the resisting capacity of the civil infrastructures against the disaster, even result in collapse. Thus, techniques for structural health monitoring (SHM) in engineering infrastructures are indispensably used to ensure safety, reduce the cost of maintenance, and prolong the service life. In recent years, Global Positioning System (GPS) has been widely used in the health monitoring of large structures since it is weather independent, capable of autonomous operation, and also does not require a line-of-sight between target points. The thesis focuses on some key technology of structural health monitoring using GPS and the following aspects are devoted to the main effects:(1) The stationary wavelet transform is applied to the wind velocity analysis. Comparing statistic analysis of wavelet coefficients of measured wind velocity and simulated data obtained by conventional harmonic wave superimposing method, the results show that the conventional method cannot reflect the local self-similarity and intermittency of fluctuating wind velocity. The wavelet scalogram and wavelet coherence analysis of wind velocity and pressures are carried on by continuous wavelet transform. The instantaneous time-frequency characteristics of them are analyzed and the results indicate that the wind velocity and pressure appears complex non-stationary and highly intermittent. For the pseudo-coherence induced by noise in wavelet coherence analysis, a statistical significance level estimation method based on Monte Carlo simulation is presented to isolate the meaningful coherence from the spurious noise map. A novel wind velocity simulating method based on inverse stationary wavelet transform is presented by multi-scale reconstruction of wavelet coefficients. In the simulation process, the Log-Poisson law is applied to generate the wavelet coefficients and the powers of wavelets at each scale are determined by a power spectrum which we selected. The results from statistic analysis of simulated data demonstrated the validity of the presented method and the wavelet coefficient sequences need not interpolated at each staged proved the convenient of the method.(2) An integrated method based on Adaptive Noise Cancelling (ANC) principles and wavelet de-noise method is presented to derive a relatively noise-free time series from the...