Research On Multipath Error Processing Method Of GPS Dynamic Deformation Monitoring

Deformation monitoring of large-scale structure of the building is the Necessary means to grasp the Security and stability during the construction and operation stage. With a high positioning precision and degree of automation, all-weather work, without the limit of intervisibility conditions and other advantages,GPS has been one of the most advanced, and widely used means of deformation monitoring. In deformation monitoring, the baseline between reference station and rover station is usually short, its public errors such as ionosphere errors can be eliminated through differencial technique effectively, but the monitoring station’s relevance of multiple is very weak and can’t be eliminated through differencial technique, so the multipath effect is the pivotal factor that restrict the GPS positioning accuracy in deformation monitoring. To solve this problem, the study of this paper mainly focused on the following aspects:(1) Research the mechanism of the generation of GPS multipath effects, analysis the characteristics such as amplitude and frequency on GPS carrier phase multipath effect.(2) Comparing the difference between several calculation methods of the repetition period of the multipath effects, experimental result shows that the repetition period of the multi-path effects calculated by the methods of maximum cross-correlation coefficient,least RMS and broadcast ephemeris have a consistency, and use this obtained repetition period to sidereal filter is better than the standard sidereal repetition period.(3) During the high-rate GPS Dynamic deformation monitoring, strong temporal correlation noises exists in the multipath effects errors, for this situation it will be not able to achieve the optimal result with the standard Kalman filter. This paper introduces an estimation method of correlated observation noise function model based on the first order Gauss-Markov process, and derives two modified Kalman filtering methods based on this noise function model, state vector expansion method and adjacent time differencing method. Processing a GPS experimental data based on provided methods, The numerical results verify the effectiveness of the proposed noise function model estimation method, both of these two methods can weak the effect of multipath effect, improve the positioning accuracy.(4) Proposes an integrated noise correction method based on wavelet filtering and the Principal Component Analysis (PCA), which select appropriate wavelet basis and decomposition level to decompose the data on multi scale, and process the high-frequency part with threshold to diminish the high-frequency random noise, then use PCA to extract and eliminate the high-correlation multipath effects. Experiment data shows that the combined method can effectively weaken the multipath effect and high frequency random noise, and is superior to single filtering method.28figures,12tables and61references.