Satellite Orbit Navigation Based On Sun-Earth-Star Information

Autonomous survival of satellite is the important field of Space-based system in 21st century.Autonomous Navigation Techniques is one of key technologies in autonomous survival of satellite.This article is aim at the Earth pointing Three Axis Stabilized Satellite, discussing the Autonomous Navigation Techniques based on the Optical Information of Sun-Earth-Star.The main research content are summarized as follows:A precise orbit dynamic model of a satellite is established in the equatorial geocentric inertial coordinate system,taking account of the perturbation of the earth's shape,the gravitation of the sun and moon,the tension of the sun light and the resistance of the air.This article also gives out measurement model of attitude sensors.An autonomous satellite navigation by the starlight angle is described.Star sensor with high accuracy and infrared earth sensor is used to calculate the starlight angle,so the orbit of the satellite can be accurately determine.After this,a new navigation concepts based on the Optical Information of Sun-Earth-Star is introduced.Simulation and analysis results showed that this model can integrate the redundant positioning information and is a good scheme for improving the positioning accuracy and robustness of the positioning system.We know shadow of the sun is a common phenomena for satellites in orbit,so a continual navigation system is simulated where positioning information changes form Sun-Earth-Star to Earth-Star.Based on Multi-sensor Information Fusion in satellite Positioning Detemination System,this article choose centralized filter and federal filer.Optimal information fusion theory tells us that choose the trace of covariance matrix as distribution factor in Federal Filter adaptive filtering. The positioning system consists of three filers-a master filter and two slave filters.The slave filters do the job of filtering,the master filter is used to calculate the positioning information from the results given by the slave filters.BP neural network can be use to satellites Autonomous NavigationAccording to orbit dynamic model of a satellite and measurement models of attitude sensors,orbit state parameters and measuring information are simulated.First,simulated data should be preprocessed.Second,design the BP neural network including number of hidden layer,node number of hidden later,training function and so on.Third, measuring information should be network inputs, orbit state parameters are outputs,training the BP neural network.Finally,we can use BP neural network to satellite autonomous navigation.The feasibility is verified via data simulation experiments.