Research On Cubaure Kalman Filter And Its Application In Multi-UAV Distributed Target Tracking

With the development of science and technology,the requirement for target tracking accuracy has grown higher and higher.Nonlinear filter methods are the basis for achieving high-precision target tracking,and have received extensive attention and researches.Cubature Kalman Filter(CKF)is the latest nonlinear filter method proposed in recent years.Adopting the third-order spherical-radial rule,it effectively overcomes the shortcomings of other nonlinear filter methods and has become a research focus in nonlinear filter methods due to its rigorous mathematical derivation,simple design,and few adjustment parameters.This thesis focuses on some problems in the target tracking application of CKF algorithm.The main research work and innovations of this thesis are as follows:(1)Aiming at the problem that single-fading factor strong tracking filter can only achieve first-order Taylor expansion precision while CKF has a second-order approximation accuracy for arbitrary nonlinear functions,which leads to inconsistent approximate accuracy,a new single fading factor strong tracking method is proposed.Firstly,a general single fading factor strong tracking nonlinear filter method is established.Then the equivalent description of the single fading factor strong tracking idea in CKF algorithm is derived,and the consistency of the approximate accuracy between the proposed single fading factor strong tracking method and CKF is proved.Finally,an improved single fading factor strong tracking CKF algorithm is proposed.By improving the introduction location of the fading factor,the fast strong tracking CKF method is proposed to reduce the time complexity of the algorithm without affecting the filter accuracy.Simulation results show that the improved single-fading factor strong tracking CKF algorithm and fast strong tracking CKF algorithm are better than the existing strong tracking CKF algorithm in terms of algorithm time complexity and filter performance,and have better robustness.(2)Aiming at the problem that current multi-fading factor strong tracking methods only consider the diagonal elements of the covariance matrix,and does not make full use of the correlation between the system states,an improved fading factor matrix strong tracking CKF algorithm is proposed,which reflects the correlation between system states in the fading factor matrix.Compared with traditional strong tracking methods,the improved fading factor matrix strong tracking CKF does not require the fading factor to be greater than 1,which avoids possible over-regulation.Then the stability and algorithm performance analysis are conducted for the two improved strong tracking CKF.Simulation results show that compared with the existing multi-fading factors strong tracking CKF algorithm,the improved fading factor matrix strong tracking CKF algorithm proposed can reduce the influence of the fading factor on the filter accuracy of the normal system state,and has better filter performance.(3)For the problem that information CKF relies on the measurement function Jacobian matrix,the influence of Jacobian matrix established by three methods on the information CKF algorithm is analyzed and compared,and the selection method of Jacobian matrix is proposed,then,an improved information CKF algorithm is proposed to extract the additional information contained in the predicted measurement.Aiming at the problem that existing distributed consensus algorithm requires member to know the global information such as the total number of vertices in the network,a novel average consensus protocol is designed to dynamically estimate the total number of vertices in the network in real time,so that the vertices leave the network or adding new vertices are easier to accomplish.For the problem that estimating the total number of vertices depends on special vertices,a maximum consensus protocol is designed,which enables the network to distributedly choose new special vertex.Aiming at the problem that the information CKF is not guaranteed to be equivalent to CKF,a distributed consensus CKF based on improved convex combination is proposed.Simulation results show that each vertex can correctly estimate the total number of vertices of the network using the average consensus protocol.The distributed consensus CKF algorithm based on improved convex combination and the distributed consensus CKF algorithm can reach the same fusion precision as the centralized CKF algorithm,moreover,when the information CKF and CKF are not equivalent,the distributed consensus CKF algorithm based on improved convex combination has better fusion precision.(4)The improved single fading factor strong tracking CKF and the fading factor matrix strong tracking CKF are applied to the UAV target tracking to improve the tracking accuracy when the target maneuvers or the maneuvering target model are set unreasonably.The improved distributed consensus CKF algorithm is applied to multi-UAV distributed target tracking,so that the UAV does not depend on any global information,and only needs to communicate with the neighbor UAV to achieve distributed information fusion structure and obtain the same fusion performance as the centralized information fusion structure,which verifies the feasibility and effectiveness of the multi-UAV vertex number estimation method and the proposed special vertex selection method.