Research On Autonomous Navigation Method For Satellite Formation Flying Based On X-ray Pulsar

The formation satellites have the advantages of good performance,low cost,high flexibility and strong fault tolerance.They are widely used in the field of near earth survey,scientific experiment,deep space exploration and so on.Autonomous navigation technology is the key to achieve cooperative autonomous operation of formation satellite.X-ray pulsar navigation(XNAV)has the characteristics of complete information,high precision and autonomy.It can be used not only as the assistant of near earth space,such as satellite navigation system,but also can provide the absolute time and spatial reference for deep space probe.XNAV can achieve full of high precision seamless navigation services for near earth space,deep space and interplanetary mission,it is an effective way to solve the problem of spacecraft autonomous navigation.At present,the study on autonomous navigation technology based on X-ray pulsar has been widely carried out at home and abroad for satellite formation flying,but the accuracy of phase estimation,initial and real-time orbit determination of the formation satellite can not completely meet the needs of the aerospace.In this paper,we mainly focus on the method of phase estimation,initial orbit determination,real-time state estimation of formation satellite,and the XNAV is applied to autonomous navigation for the Mars Surveyor.The main contribution of this dissertation can be outlined as follows.(1)A weighted direct phase estimation algorithm in frequency domain for photon sequence is proposed to solve the problem that the error of accumulated phase estimation in the average profile is too large for X-ray pulsar signals.The photon sequence that is captured by the X-ray detector and then sampled is directly transfromed to the frequency domain,then the phase of each frequecy is respectively extracted and is weighted according the amplitude of the frequency component.Furthermore,the segmented FFT strategy is adopted to solve the memory overflow problem caused by long sequences.Finally,the computational complexity of the proposed estimator is investigated in theory.The simulations and comparisons with existing methods under the condition of the same observation time and sampling interval show that the accuracy of the weighted approach increases by 10% over the average method and is two times higher than those of the Nonlinear Least Sequare(NLS)and Maximum Likelihood(ML)methods,and that the computational complexity of the algorithm is between NLS and ML methods.Moreover,the performance of the algorithm with segments is improved by increasing the number of segments.(2)In view of the problem that the spacecraft needs to be fitted with multiple detectors and the low precision of initial orbit determination,a method using only one detector observes three pulsars in turn to determine the initial orbit of satellite is presented in this work.To improve the performance,the use of the incremental phase in one observation duration is proposed,and the incremental phase is combined with the time difference of arrival(TDOA).Then,weighted least squares(WLS)algorithm is formulated to calculate the initial orbit.Numerical simulations indicate that the method combined with the phase increment observation proposed in this paper gives better performance than that without the phase increment.The precision of position is improved by more than 27% and the velocity by more than 18% with the new method.This initial orbit determination precision of this new method meets the rough initial determination requirement.(3)Aiming at the problem of low accuracy of real-time positioning in the spacecraft formation flying navigation,an X-ray pulsar navigation method based on the Unsented Kalman Filter(UKF)with the information included angle between the baseline vectors,incremental phase and timing observation is proposed.Firstly,the geometric model of the baseline vector under the circumstance of two satellites and three satellites formation flying is introduced.Then combined with the satellite orbit dynamics model,the corresponding UKF filter model is designed according to the established mathematical model;Finally,the navigation performances of many combinations are verified.Numerical simulation results indicate that the navigation precision obtained by the new method can be improved by more than 12% in position and 9% in velocity estimation in two satellites formation,while more than 21% in position and 12 % in velocity estimation in three satellites formation.The agumented method increases the amount of information available effectively,reduces the state estimation error,and improves the convergence speed and the precision of navigation to some extent.The performance is better than that of timing and vector observation method individually,which demonstrates the enhanced XNAV algorithm is feasible and effective.(4)An augmented XNAV method based on the field angle and natural satellites' direction vectors of Mars is proposed to solve the problem of poor autonomy and low accuracy for Mars probe spacecraft.The measurement model of field angle and direction vectors is formulated by processing satellite images of Mars obtained from optical camera.Then,the filtering model is established by intergrating the observation model and the X-ray pulsar timing observation model into the Mars probe orbit dynamics equation.In view of the invisibility of Mars' natural satellites in some cases,a visibility condition analysis is given.Finally,the performance of two different orbits state estimation is simulated.In order to improve the accuracy of state estimation and reduce the impact of the system noise on navigation precision,an adaptive divided difference filter(ADDF)is applied.A set of comparison experiments of ADDF with UKF,DDF and EKF under the same condition of visibility are carried out and the results show that the ADDF has fast convergence speed,strong adaptability to noise and the state estimation performance is improved by 40% compared with UKF,DDF and EKF.In order to verify the validity of augmented XNAV positioning system,numerical simulation are carried out in three cases.In the first case,none of the Mars' satellites are visible.In the second case,only one satellite of the Mars available,and in the last case,both of the satellites of Mars are visible.Numerical simulation results show that the positioning precision is the highest in the last case.Moreover,the navigation precision is evidently improved more than 20% by using the augmented XNAV in a comparison with the conventional XNAV.