A Study On Some Processing Methods For Survey Data Based On Virtual Observational Approach

Semi-parametric Regression, Vondrak Filter and Ridge Estimation are the three main approaches in the data processing of Geodetic. There are parametric components and nonparametric components included in the Semi-parametric Regression model, which more reflect the true world and make good use of the information when used to describe the problems. Vondrak Filter can be used to cope with the data serials with time influenced by the high frequency noisy. Ridge Estimation is used to deal with the ill-condition problems in the data processing.There is a common problem to determine the smoothing factor or ridge parameter. Whether the factor is suitable or not influences the quality of the data processing directly. Hoverer, there are many disadvantages in the way to determine the smoothing factors at presents.The thesis based on the former research, from the principle of the Semi-parametric Regression, Vondrak Filter and Ridge Estimation, proposed a new approach to determine the something factor or ridge parameter based on the virtual observational approach. The key of this method is to make the smoothing factor as the weight coefficient by using virtual observations theory, the Helmert type variance component estimation then was used to find the optimal factor. When the ratio of variance of fitness to smoothing equal to one or through the certain way to consider that the ratio equal, then the factor is the optimal. The main contents as follows:(1) The thesis proposed Semi-parametric Regression based on the virtual observational approach, discussed the relationship between the new approach and the traditional approaches. It is used to solve the deformation monitoring problem and gets good results.(2) The thesis proposed Vondrak Fileter based on the virtual observational approach. Through analysing the simulation data and real experiment data, proving that the new method is feasible, using the new approach to mitigate the GPS multipath effects and get good results. Thesis also discussed the relationship between the new approach and the Penalized Least Squares. (3) The thesis proposed Ridge Estimation based on the virtual observational approach, and getting that the new approach is feasible through comparing with the others.