Research On Quasi Geoid Refinement Method Based On Grid Model

In recent years,due to the continuous in-depth research and development of GNSS navigation and positioning system,it has the advantages of three-dimensional positioning and speed measurement,accurate timing,real-time all-weather,free resources and so on,and has been widely used in the field of Surveying and mapping.High precision quasi geoid results can make GNSS directly determine the normal height,give full play to its advantages in the field of Surveying and mapping,improve the efficiency of operation,and bring direct economic benefits.It is of great practical value to study high-precision quasi geoid,which has become one of hot topics for surveying and mapping scholars.The subject of quasi geoid refinement is studied in this paper.The theory and model of RBF(Radical Basis Function)neural network and its application in height anomaly fitting are introduced in this paper.In view of the shortcomings of the fitting method of quasi geoid refinement in large areas,a quasi geoid refinement method based on grid model is proposed.Six quasi geoid models are established in this paper,and the accuracy of each model is verified by the data of examples..The main contents and achievements of this paper include the following aspects:(1)The theory and methods of quasi geoid refinement commonly used in engineering are introduced in this paper,and the geometric method,gravity field model method and combination method are analyzed in detail.Firstly,the quadratic polynomial model of geometric method is used to fit the height anomaly,and the EGM2008 gravity anomaly data is added to establish the quadratic polynomial quasi geoid model with gravity information;(2)The RBF neural network constructed by using Gaussian function as the activation function of the hidden layer of the neural network shows good effect in the process of height anomaly fitting.The new model is built by adding egm2008 gravity anomaly data in the input layer of the model.The experimental results of two engineering examples show that the accuracy of the model is improved on the basis of RBF neural network model fitting results.(3)In view of the shortcomings of the partition fitting method,a seamless partition idea is proposed,which is a large area quasi geoid refinement method based on grid model.The results show that the accuracy of grid model is better than that of the conventional quadratic polynomial model and RBF neural network model;The improved grid model is used to perform elevation anomaly with EGM2008 gravity data as known parameters,and the fitting accuracy is higher than the other five models.The results of two engineering examples show that the improved RBF neural network model and the improved grid model are both of good precision,which can be complementary to each other and suitable for practical engineering application.