GNSS Surface Elevation Fitting Method Based on Machine Learning Optimization
To address the phenomenon that the traditional surface fitting method and the nonlinear machine learning algorithm are frag-mented in the process of normal height conversion of geodetic heights in the region. This paper proposes to optimize the linear surface fitting method by using BP neural network and random forest ( RF) algorithm,and conducts experiments by using the measured GNSS and level data from a mining observation station. The experimental results show that the nonlinear optimization of the surface fitting results using BP neural network can achieve more accurate GNSS elevation fitting accuracy in the case of large observation area and anomalous elevation irregularities. It provides an idea for further improving the accuracy of ground elevation observation,which is of practical significance for various engineering construction and mine deformation monitoring.
BP neural networkRF algorithmarea modelingelevation fitting
NETL
NSTL
万方数据
