Research On Tracking Techniques Of Maneuvering Group Targets

Group target tracking has been a research focus in the field of aerospace in recent years.Due to the complexity and diversity of its maneuvering modes,the number of individuals in the group is large,and the overlapping phenomenon is serious,the traditional multi-target algorithms cannot track the group effectively.In addition,it is of great significance to cognize situation of the group by estimating its extensive state such as size,shape and direction.In this paper,the centroid group target tracking algorithm and the Bayesian data association algorithm are studied for “the uncertainty of target motion mode” and “the uncertainty of the corresponding relation between measurements and targets”.The key techniques about the group splitting and merging based on random matrix are researched considering the special role of group extensive state in group target tracking.The major contributions are given as follows:1.A new algorithm of centroid group target filtering based on integrated adaptive interactive multi-model is proposed aiming at “the uncertainty of target motion mode”.The algorithm mainly addresses the problem of the high computational complexity and fixed model transition probability in the traditional multi-model algorithm,such as poor real-time and large errors.Firstly,the output value of the interactor is weighted and integrated,so that each sub-filter selects the same candidate echo instead using different measurement sets for status update,thus reducing the redundancy calculation.Secondly,the Bayesian theory and the model probability change rate are used to increase the probability of superior model and reduce the adverse competition between models,to increase tracking accuracy.Finally,the simulation experiments prove the validity of the algorithm.2.A multi-feature information fusion algorithm for group target data association with adaptive gate is proposed aiming at “the uncertainty of the corresponding relation between measurements and targets”.Firstly,the correlation between candidate echoes and true targets is established through targets' electromagnetic radiation characteristic information and motion state information to achieve the accurate interconnection of measurement.Secondly,in view of mistaken tracking or losing targets caused by the reduction of effective measurement within the association gate during formation maneuvering,an adaptive association gate which can be modified at the time of maneuvering is designed by innovation covariance function.Finally,the simulation results demonstrate that the proposed algorithm is superior to traditional methods in tracking accuracy and loss rate.3.A group splitting and merging detection and tracking algorithm based on extensive state estimation algorithm is proposed aiming at the problem that a single ellipse cannot to be used to describe the group shape after splitting or merging in the traditional random matrix algorithm.In the group splitting scenario,the random matrix is used to estimate the change rate of the extensive state to detect the situation of the group in real time.When group splitting is detected,k-means clustering is used to divide the separated measurements into several smaller groups,and the least square method is used to synthesize the groups into ellipses for real-time estimation of the extensive state.Group merging is equivalent to the inverse process of splitting.Simulation results show that the proposed method realizes the real-time estimation of the extended state after group splitting and merging and overcomes the problem of slow convergence of the random matrix algorithm.