Research On The Relative Navigaition Of Formation Flying Satellites Based On Adaptive Kalman Filter

With the increasing strategic position of space science technology,satellite formation flying aimed at expanding system capabilities has been paid more and more attention.The performance of many key technologies such as high-precision formation control of formation satellite systems heavily depends on high-precision relative navigation.Therefore,it is of great research value and significance to analyze the characteristics of relative navigation problems in different formation scenes,to study the filtering method that can accurately and efficiently determine the relative position and velocity between satellites,and to improve and optimize the relative navigation performance of formation satellites.For applications to relative navigation of formation flying satellites,considering the influence of different reference orbits of formation flying satellites on relative navigation performance,this dissertation describes the filtering problems with different motion models and noise/residual characteristics,and proposes problem-specific schemes based adaptive filtering methods to improve the accuracy,efficiency and adaptability of relative navigation.Main research and innovations are given as follows:Firstly,a new adaptive filtering algorithm with feedback adaption of prior error covariance is proposed for the relative navigation problem of circular orbiting satellites,which is modeled linear time-invariant system with inaccurate statistical parameters for process noise.Simulation and comparison results demonstrate new adaptive algorithm’s advantages of better estimation accuracy,lower computational complexity and superior adaptability to filtering cases of inaccurate process noise parameters.Secondly,to cope with unpredictable approximation errors caused by the discretization/linearization of nonlinear continuous dynamic model of near-circular orbit formation satellites,a nonlinear continuous-discrete adaptive filtering algorithm is proposed based on the novel online feeback scheme with posterior stochastic sequence.Simulation results show that this approach can more efficiently and effectively suppress the influences of discretization/linearization errors on filtering accuracy and stability.Thirdly,to deal with unacceptable model residuals introduced during cubature rule based transformation and disretization to strongly nonlinear continuous dynamics and nonlinear discrete measurement model,a new continuous-discrete version of cubature Kalman filter based on the NIRK integration of cubature sampling vector and adaptive error regulation is proposed for the relative navigation of elliptic formation satellites with large eccentricity.Simulation results demostrate that the adaptive method can effectively regulate the magnitude of nonlinear transformation and discretization residuals,which can in turn improve the new method’s estimation accuracy and filtering stability.In summary,the relative navigation of circular,near-circular and large eccentricity orbitting formation satellites are constructed as three filtering problems for linear time invariant system,nonlinear continuous dynamics and strongly nonlinear continuous dynamics/ nonlinear discrete measurement system.Considering the different characteristics of process noise or approximate errors of these filtering models,this dissertation puts forward corresponding new adaptive Kalman filters.The simulation results of typical formation scenarios verify the filtering performance of the new adaptive algorithms,which is expected to provide theoretical basis and experience for the research and application of related filed.