Orbit Determination Of Non-cooperative Targets Based On Binocular Vision

In today's world,space situational awareness has become the technical innovation field of various countries,and non-cooperative target detection technology has become the focus of research and development in various countries.Non-cooperative target orbit determination is an important technical means to determine the real-time position of non-cooperative targets in space.There are many detection studies,but deep space and long-distance non-cooperative target detection is still at an immature stage,with few studies and insufficient innovation.The thesis analyzes the advantages and disadvantages of each method and improves it through non-cooperative target detection and orbit determination methods and domestic and international orbit improvement algorithms.In-depth research and simulation analysis are made on the problems of non-cooperative target deep space detection methods and detection system structural error improvement,initial orbit determination model design,precise orbit determination algorithms,and orbit determination error propagation.Based on the detection methods of non-cooperative targets,the scope of application of ground-based and space-based detection is compared,and the advantages of space-based optical detection in space target detection are explained.By comparing with the single-eye detection and multi-eye detection of traditional optical detection systems,the analysis The advantages of the binocular detection system with small load,high detection accuracy,and depth information for non-cooperative targets can be obtained.In order to achieve the ability of the binocular detection system to detect far-reaching targets in space,it is innovative to put an optical camera on two satellites in the same orbit to form a binocular detection system.By increasing the baseline distance of the binocular detection system,it achieves deep The function of target detection.In order to reduce the structural error of the binocular detection system,on the basis of the established accuracy analysis model,the structural parameters of the binocular detection system are optimized byimproving the particle swarm algorithm.The simulation results show that this intelligence is applied when the detection depth is determined The binocular detection system designed by the optimized method can effectively reduce the detection error.A GUI interactive interface is used to establish the initial orbit determination model of the non-cooperative target.The depth information obtained by the detection system is obtained by the Newton iterative algorithm and the initial orbit is determined by the traditional r-v orbit determination method.The determination of the initial orbit is rough and the Newton iteration takes a long time,and it is impossible to obtain high-precision orbit information of non-cooperative targets in a short time.Innovatively proposed a joint improved algorithm of neural network and unscented Kalman filtering.Through neural network training and detection of data,a neural network state transition model was established.Through the combination of this state model and unscented Kalman algorithm,the improvement of unknown non-cooperative target orbit the goal of.This method solves the bottleneck of error optimization of the non-cooperative target unknown state model,and provides a new way of improving the track.Simulation analysis shows that this method is beneficial to the acquisition of 3D information of deep space targets,and effectively improves the accuracy of non-cooperative target orbit determination.Through simulation analysis,this joint orbit determination method can reduce the maximum orbit determination error from 40 meters to under 5 meters.The thesis uses a combination of theoretical derivation and numerical simulation analysis,focusing on issues such as non-cooperative target detection and non-cooperative target orbit orbit improvement.Such research has important reference value.