| A lot of disasters are related to the deformation of man-made or nature structures, so the deformation monitoring is very important. The monitoring results can help to check the design of the structures, assess the health conditions and prevent disasters due to structural failures. The traditional sensors for large structure monitoring such as accelerometer can not measure the static or low frequency part of the displacement. This part is the inertial whole displacement of the structures duo the force such as the typhoon. It is very important for structural health assessment. GPS can measure the absolute displacement and has many advantages such as no weather condition limited, high automatic etc. So GPS has been increasingly used in monitoring the dynamic behaviors of large structures such as high-rise buildings and long suspension bridges especially under abnormal loading conditions such as during earthquakes and typhoons. However the accuracy of GPS can not be satisfied the precise application such high-rise building monitoring. In additional, there will be a lot cycle slips in monitoring environment for GPS. And the RTK GPS will become unstable, because it needs some time for initialization when there is cycle slip. In order to improve the accuracy and the stability of GPS in dynamic deformation monitoring, the GPS stochastic model, residual tropspheric delay, multipath and single epoch ambiguity resolution method are discussed in this thesis. Following are the main contents of the thesis:1) Several GPS stochastic models based on quality indicator such as satellite elevation angle, carrier phase to noise ratio and signal strength etc are compared. The results show that the GPS stochastic model based on C/No or signal strength is very helpful to mitigate diffraction errors. And the stochastic model based on satellite elevation is more efficient to mitigate the residual tropospheric delay errors.2) A robust GPS stochastic model combining C/No and priori information was developed. This model is more effective to mitigate the unmodeled system errors such as diffration.3) Theoretical and experimental studies have shown that tropospheric effects can become a significant error source when the height difference is large. A residual zenith tropospheric delay estimation method considering the stochastic model and priori information was developed. This method is effective when the residual tropspheric delay error is close to the multipath or diffraction.4) The character of multipath in dynamic deformation monitoring was analyzed from both theoretical and experimental aspects. The results show the repeatability of multipath will be affected since the distance between reflector and antenna has been changed. Fortunately, the low frequency part of the multipath in dynamic deformation monitoring still repeats daily to some extent.5) A noise filter (EMD noise filter hereinafter) based on the characteristic of IMF (intrinsic mode function) is developed. Compared with wavelet method, it is more straightforward, and not subject to the effects from the selection of certain wavelet function. It therefore can better extract and thus model the repeatability of the multipath errors, which is crucial in mitigating the errors.6) A GPS single epoch solution algorithm with constraints of deformation characteristics was developed. The Cholesky decomposition method was used to set up the ambiguity search space directly which become very small if there is only horizontal or vertical deformation. From the GPS monitoring test, it shows that the method is very efficient, and the success rate is also improved much more.7) At last, a GPS data processing software for precise dynamic deformation monitoring was developed using visual C++. The real GPS data for monitoring a residential high-rise building in Damry typhoon condition was processed using this software. The result shows that GPS is very useful in structure health monitoring.
|
PDF Full Text Request