Multiple Resolvable Groups Tracking By Using Random Finite Set

Group targets consist of a number of cooperative targets and move in a certain structure.With the sustainable development of science and technology,the original unresolved group targets can be oberserved by sensors with a resolvable attribute.The research on resolvable group target tracking is increasingly important.The tracking estimation mainly includes obtaining the measurements of targets,estimating the states of the targets from the measurements.Therofere,it is necessary to establish the dynamic model of the group targets and find a suitable algorithm to track the group targets.At present,the dynamic modeling of the group targets is just a simple combination of the state of each member in the group,but this method can not fully reflect the dependent relationship among the members.The current research on group target tracking mainly focus on unresolved or partly resolvable group targets.However,the tracking approach of the unresolved group targets can not be applied to the resolvable group target,for it is too rough and loses structural information.On the basis of graphs theory and labeled random finite set(L-RFS),a resolvable coefficient is first introduced.Then,dynamic modeling and the tracking estimation of group targets are proposed.The whole structure of the thesis is as(1)Dynamic modeling of resolvable group targets.According to the similarity between group structures and graph structures,the group is modeled by directed graph.The member dependency is determined by an adjacency matrix.The leading node plays a main role and is modeled separately.Then,the dynamic models of other children nodes are built.(2)Resolvable coefficient of the group targets.The radar resolution decreases with the increase of the detection range,which will leads to the worse of the detection performance.Assuming that the radar detection error obeys normal distribution,the 3?distance standard between the mean of the normal distribution and the edge is adopted.Therefore,in this paper,the resolvable distance of the group target is3 3i j?(10)?,where i and j represent two targets,respectively.When the distance between individual pairs of members in the group target is greater than certain resolvable coefficienti,jr,the group target is considered to be resolvable.(3)Research on the tracking algorithm of the resolvable group targets.Since the collaboration between groups is unknown at the begining,preliminarily assuming the independence among members of group targets is reasonable.Meanwhile,the state,trajectory,and the number of targets estimation are given through the GLMB algorithm.Besides,based on the state estimation of each member in the group targets,the deviation matrix and the adjacency matrix are calculated at each scan.Finally,the structural relations is analyzed from the adjacency matrix.The number of subgroups is estimated by using the concept of connected graphs.