A Study Of Multi-Scale Analysis For Dynamic Deformation Monitoring Using GPS

Compared with traditional instruments, GPS has many advantages such as no weather condition and line-of-sight limited. The most important advantage is that GPS can measure both the high frequency and quasi-static displacement of large engineering structure under abnormal loading conditions such as during wind and temperature etc. So GPS has been increasingly used in structure health monitoring. However the GPS measurements include not only the real displacements of the structures, but also the quasi-system errors such as multipath and residual tropospheric delay. In order to filter the noise, mitigate the quasi-system errors and analyze the displacements, two multi-scale analysis methods, i.e. wavelet analysis and empirical mode decomposition are used in using GPS dynamic defomation monitoring. Following are the main contents of the thesis:1. Data characters analyse of dynamic deformation monitoring using GPS. It includes non-linearity, non-stationary, complex periodicity deformation, noises, system error and et al.2. The basic theories of muti-scale analysis, wavelet transform and empirical mode decomposition. The continue wavelet transform, discrete wavelet transform and empirical mode decomposition were discussed first. Then the basic theories of muti-scale analysis were discussed in detail.3. The multi-scale analysis method based on wavelet and its application in dynamic deformation monitoring. It includes denoising, data analysis in time-frequency domain, and extracting the properties of the dynamic deformations.4. The multi-scale analysis method based on empirical mode decomposition and its application in dynamic deformation monitoring. Simulated experiments and real GPS data were analyzed using empirical mode decomposition. The corresponding results were compared with the ones from wavelet method .5. A new multi-analysis method based on the combination of wavelet and empirical mode decomposition was developed. This method can overcome the disturb of noise using empirical mode decomposition, reduce the boundary effect and decomposition number, and improve the accuracy of the extracted properties of deformation.