The Theoretical Research Of Kinematics Positioning Of GPS

Based on the current status and existing problem of the highly accurate kinematic positioning of GPS, this dissertation systematically studies the theory and method of highly accurate kinematic positioning of GPS. This study centers on utilizing a prior information to improve the resolution of integer ambiguity.In real-time kinematic (RTK) GPS measurement, integer ambiguity resolution is always very difficult, especially in the case of long baseline. Integer ambiguity resolution and cycle clip detection and repair are the crucial problems in RTK-GPS positioning. This dissertation includes seven parts:First, the methods of integer ambiguity resolution are summarized.Second, the characteristics of different error sources in GPS observable are analyzed and the measures that should be taken are discussed.Third, the method to improve ambiguity resolution with geometry constraints is presented. At the same time, the principles of the method, the relationship between the values of geometry constraints and the searching space of ambiguities are; also described. A method to determine the value ofgeometry constraints is singled out.Fourth, the quasi-observable method is explained. A simple algorithm utilizing the quasi-observable to conduct the adjustment is presented and the capability using quasi-observable method to improve the ambiguity searching is studied in detail. The results show that quasi-observable method can not only enhance the residue test, but also optimize the structure of searching space. Having discussed the reliabilities of quasi-observable method, this dissertation further presents a method to weigh the quasi-observable. Field tests demonstrate the effectiveness of those presented methods.Fifth, the application of Bayes method in GPS positioning is discussed. In this part, the traditional solution of the least squares adjustment with inequality constraint (LSI) is analyzed at first. Then based on uniform prior distribution, Bayesian solution of LSI is proposed. The results show that traditional solution of LSI is actually a Bayes solution based on the mode of the posterior distribution and that it is not the best solution in the sense of variance minimizing. In most cases, Bayes solution based on the mean of the posterior distribution will be better than the traditional one. At last the Bayesian algorithm of LSI is applied to GPS positioning. Results from four field tests show that when height information is used, the GPS ambiguity resolution can beimproved significantly by applying Bayesian algorithm to LSI.Sixth, the relationship between phase observables of the two carrier frequencies is analyzed. The fast integer ambiguity resolution method is studied using phase observables of the dual-frequency carrier in combination with C/A code.Finally, the method utilizing double differentiation of phase ovservables of neighbor epochs to detect and repair cycle slip is presented. The reliability of the proposed method is demonstrated by experiment.