Fusion Tracking Algorithm Research Of Maneuvering Target

In the field of modern military and civil fields, maneuvering target tracking occupies an important position and has a broad application prospect,which attracts many researchers at home and abroad launching a deep research and make the leap development. Maneuvering Target Tracking is a study of the mobile target estimation problem which can not describe the target maneuverability accurately. With the rapid development of science and technology, higher requirements for maneuvering target tracking are put forward because of the higher standard of target speed and mobility in different complex tracking environment. In the case on the basis of predecessors’research results and the combination of latest developments at home and abroad, This dissertation is concerned with the control problems deeply studied with state estimation and interactive multiple model (IMM) algorithm.First of all, this paper introduce the background and research state in both overseas and domestic in the theory of maneuvering target tracking,it also involves both target tracking and the mathematical model of target motion and sketches out the basic principle of maneuvering target tracking. Also, the maneuvering target tracking state estimation and Kalman filtering algorithm is proposed with the theoretical derivation and the detailed process of the algorithm. We then analyzed the theory of probability data association and Kalman filtering algorithm, as well as gave the concrete steps of the algorithm.Secondly, under the background of maneuvering target tracking, we presented nonlinear filtering theory and EKF, UKF and CKF three kinds of nonlinear filtering algorithms that were elaborated their the theoretical foundation and basic principles associated with the detailed process. The author deduced a nonlinear target tracking model to make a comparison and verify the tracking performance ranging three algorithms.we highlighted the theoretical foundation and implementation methods and specific steps of particle filter(PF). The new ICKPF is based on the Markov Chain Monte Carlo(MCMC) and uses the cubature rule based on numerical integration method to calculate the mean and covariance, which generates the proposal distribution for the particle filter. The proposed algorithm involves the current measurements in the proposal distribution so that the degree of approximation to the system posterior density is improved. Meanwhile, the algorithm is optimized by MCMC sampling method, which makes the particles more diverse.Eventually, according to the standard of the interacting multiple model filter using Kalman filter algorithm is difficult for the tracking problem of nonlinear strong maneuvering target, the interacting multiple model was combined with the improved particle filter algorithm. An improved interactive multiple model particle filter algorithm was ladled out,and we provided the algorithm’s procedure. This method makes up the deficiency of general interactive multiple model particle filter algorithm by the simulation results verified the correctness and accuracy of the algorithm. It turned out that the proposed algorithm performed better than that interactive multiple model particle filter algorithm in maneuvering target tracking.