Mitigating the systematic errors of E-GPS leveling using neural network method

Abstract

e-GPS (Electronic Global Positioning System), which is used in Taiwan, is a type of real-time kinematic satellite positioning technology, such as VRS-RTK (Virtual Reference Station Real Time Kinematic). The height difference between ellipsoidal height h and orthometric height H is called undulation N. For a point p, if the values of h from e-GPS and N from regional geoid model are known, the H value of point p can be calculated using the following equation: H = h-N. This is the basic principle of e-GPS leveling. Data analysis of test results revealed that the estimated orthometric height from e-GPS leveling may have systematic errors. This paper proposes a BP (back-propagation) neural network and BP neural network method (BP&BP) method to manage the systematic errors of the estimated orthometric height Ĥ from e-GPS leveling. The main goal of the proposed method is to mitigate the systematic errors of orthometric height H from e-GPS leveling efficiently. Subsequently, the e-GPS leveling accuracy may be improved. Three data sets (including plane coordinates, ellipsoidal height h from static GPS, orthometric height H from first-order leveling, and ellipsoidal height h from e-GPS) of 145 benchmarks from Tainan City, Taiwan, were used to test the proposed method. The test results show that the proposed method can mitigate the systematic errors of orthometric height H from e-GPS leveling efficiently. The proposed methods and the detailed test results are presented in this paper.