Research On Relative Navigation Strategy Of Formation Satellites Based On Improved Cubature Kalman Filte

With the development of aerospace technology,formation flying has become a research hotspot in the space field.Relative navigation is one of the key technologies to realize formation flying,and its estimation accuracy and reliability directly affect the formation missions.For this reason,this paper takes the formation of near-circular orbit satellites as the application background,considering indicators such as estimation accuracy,real-time performance,and reliability.Aiming at the problems of noise change,maneuvering and high state estimation dimension in the relative navigation process,the adaptive filter and the distributed relative navigation strategy are studied.First,the relative navigation model of formation flying has been established and the common non-linear filter algorithms are introduced from the perspective of Bayesian estimation.At the same time,it is verified through simulation that the Cubature Kalman Filter(CKF)algorithm is more suitable for inter-satellite relative navigation.Secondly,aiming at the problem that the unknown statistical characteristics of the measured noise may cause the filter estimation accuracy to decrease or even diverge,an improved adaptive CKF algorithm based on noise estimator is proposed.The algorithm combines Sage-Husa adaptive filter with SVD-CKF,Singular value decomposition enhances the numerical robustness of the algorithm,and the cubature rule gets rid of the limitations of linear system applications.The improved posterior estimator avoids the negative definite phenomenon that may occur in the online estimation of noise variance.Combined with the characteristics of the formation flying motion model,the time update of the conventional Kalman Filter is used to replace the corresponding cubature transformation process,which reduces the amount of calculation without affecting the performance of the filter.The simulation results show,compared with CKF,the estimation accuracy of this method is improved and the stability is better.Then,aiming at the problem of long convergence time and low accuracy due to sudden state changes during satellite maneuver,the idea of strong tracking filtering is introduced into the SVD-CKF framework,and an improved strong tracking cubature filter algorithm is proposed.The algorithm calculates the fading factor based on the cubature change,avoiding the solution of the Jacobian matrix;through the chi-square test of the residual sequence,the fading factor is more accurate to be used;adopt the same time update and simplification steps as KF,which reduces the number of cubature points generation and reduces the computational complexity.The simulation results show,the improved method can still maintain good estimation performance during the maneuver period.Finally,in view of the single-point failure risk of centralized filtering of multi-satellite formations and the problem of excessively high state estimation dimensions,a distributed filter structure is introduced and the traditional estimation is decomposed to multi-satellite joint execution.Under this strategy,each satellite divides the measurement configuration through communication negotiation,and completes its own state estimation based on different local measurement information.In order to improve the estimation accuracy,the results of multiple local state estimations are fused.The simulation results show,compared with centralized filter,the calculation efficiency of the navigation system is guaranteed;compared with the filter of single measurement information,the accuracy of the navigation system is improved.