Research And Simulation Of Satellite Positioning Error Compensation Technology Based On Convolution Neural Network

In recent years,with the development of science and technology and the improvement of people's living standard,a growing number of location-based services appeared.Positioning technology plays an increasingly important role in the fields of military,life and scientific research.Satellite positioning technology is currently the mainstream of outdoor positioning technology,satellite positioning has high positioning accuracy,a wide range of uses,large coverage area and all-weather operations and so on.However,the signal of satellite positioning is easy to be disturbed,and there is the problem that the signal is interrupted by obstruction.At the same time,the satellite positioning has some defects such as signal instability and long signal cycle.This paper presents a satellite positioning error compensation technique to solve the defect of satellite positioning.In this paper,a satellite positioning error compensation technique based on depth learning is proposed to solve the defect of satellite positioning,and the depth learning is applied to the satellite positioning error compensation for the first time.First,by collecting large amount of data,training a good error compensation model,compensating the error of satellite positioning,improve the robustness of satellite positioning.In the course of research,this paper begins with the depth learning.In this paper,the three kinds of deep learning networks,such as convolution neural Network(CNN),depth belief network(DBN)and cyclic neural Network(RNN),are studied.After fully understanding the characteristics of three kinds of neural networks,this paper trains three kinds of neural networks using the satellite location data collected by the Institute of Technology of CAS,and uses three kinds of neural networks to establish error compensation models respectively to compensate for the error of satellite positioning.After many experiments,the error compensation model mentioned in this paper has achieved good compensation effect,effectively solved the positioning and migration of satellite positioning in the presence of signal fluctuation and occlusion,and improved the robustness of satellite positioning.In the three kinds of error compensation models proposed in this paper,the error compensation model based on convolution neural network has obtained the best compensation effect,and the compensated satellite positioning error is in the range of 1 m to 3 m,which satisfies the requirement of precise positioning.