Researches On Fusion And Data Processing For Dynamic Geodetic Measurements

The dynamic geodetic measurements include fiducial station observations, yearly repeated monitoring station observations and regional network observations, and so on. The dynamic geodetic measurements, obtained by using different designed projects, different instruments, different satellite orbits and different datums, maybe exist some systemetic errors, stochastic errors and gross errors, which probably make the estimated parameters and their posterior covariance matrix unreliable. In order to process the dynamic geodetic measurements reliably and practicablly, this dissertation presents some theoretical methods and algorithms, including adaptively sequential adjustment, compensation method of outliers, adaptive collocation. The methods of Voronoi map and minimum curvature are applied in datum transformation. The data fusion of various observations, the methods controlling outliers, the establishment of function and stochastic model, the unification of different datums, and the determination of velocity are also researched. The main works and contributions are summarized as follows:1. After analyzing the influences of the outliers in the prior parameters and measurement data on the estimated parameters of the sequential adjustment, two methods of quality control are discussed, one is the adaptively robust sequential adjustment by adjusting the stocashtic model, the other is functional compensation method by introducing some extra-parameters based on the outlier hypothesis testing, which can estimate the values of the outliers and the parameters simultaneously. An actual example is given which shows that both methodes can efficiently control the influences of outliers.2. For some special purposes, there may exist some constraints among parameters, which would affect the estimated parameters and their posterior covariance matrix. After analyzing the influences of constraints on the estimated parameters, a set of applicable formulas for sequential adjustment with constraints, adaptive sequential adjustment with constraints, and robust sequential adjustment with constraints are derived in detail, and can be used in analyzing the influences of constraints on the estimated parameters quantificationally.3. The data fusion of observations with different type and different precision is a foundational content of data process for the dynamic geodetic measurements. Several data fusion methods based on robust estimation, variance component estimation of various observations are proposed and compared. The adaptive sequential adjustment is also proposed, which can not only adjust the weights among various observations, but also adjust the weights between the observations and the prior parameter information by using the adaptive factor. Especially, the sequential algorithm based on robust variance component estimation is proposed in detail, which makes use of the advantages of robust estimation, sequential adjustment and variance component estimation. This algorithm can not only weaken the influences of outliers on the adjustment results, but also can avoid storing lots of the historical measurements and calculating the complicated matrices.4. In the applications of collocation, the prior covariance matrices between signals and observations should be consistent, otherwise, the solution of collocation will be twist. In order to balance the covariance matrices of signals and observations, an adaptive collocation estimator is derived, in which the corresponding adaptive factor is constructed by the ratio of the variance components of signals and observations. A maximum likelihood estimator of the variance components is thus derived based on the collocation functional model and stochastic model. Additionly a Helmert type estimator of the variance components for the collocation is also introduced. As the adaptive collocation is effective in balancing the contribution of observations and signals in the collocation model, the reasonable and consistent covariance matrices of signals and observations can be arrived through the adjustment with the adaptive factor.5. The unification of different datums plays an important role in data fusion of various geodetic observations, and the accuracy of datum transformation is related to the distribution and quanity of common points. After discussing similarity transformation, three methods, processing the local systematic error and distortion between datums effectively, are applied for datum transformation, one is the adaptive collocation based on statistical properties, another is the transformation based on Voronoi Map, and the third is an interplation method using the principle of minimum curature. An actual numerical example from 1980 Xi'an Coordinate System to CGCS2000 shows that the accuracy of the transformed coordinates is evidently improved by using these methodes, comparing to similarity transformation. The adaptive collocation is the most method.6. The repeated GPS network is a primary way to get the velocities of the observed stations, so the systematic errors of GPS and the compensation methods of them are discussed in detail. Two methods are presented for the determination of velocity, one is the direct method by processing baseline vectors, the other is the indirect method by using coordinates at different epochs. A new method is developed for velocity computation by introducing adaptive filter, which can not only gain datum through background field, but also blance the contribution between baseline vectors and the prior velocity from a background field.