Research On Mixed Target Estimation Problem Of Labeled Random Finite Set With Structural Relations

Mixed target are a mixture of point target,extended target and group target.In the traditional hypothesis,the mixed target groups are independent of each other and the group target distinction attributes remain unchanged,which leads to the lack of structural information of the mixed target,the increase of estimation error,and the problem of a large target missing rate.A point target in a mixed target can produce one measurement at most,while an extended target and a group target can both produce multiple measurements.The sub-targets contained in the group goal are not completely independent,and they have a certain form of cooperation,while the sub-targets contained in the extended target have no cooperative relationship.Based on the theory of RFS,a type probability algorithm based on the structural relationship is proposed,then the typed GLMB algorithm is used to track mixed target,and finally a GLMBbased target tracking estimation algorithm for long-period and large batches of mixed targets is proposed.The specific research content is as follows:(1)A GLMB-based type probability algorithm is proposed according to the structure relationship of the mixed target.Firstly,the concept of type probability is proposed,and two types of probability based on shape and coverage are introduced,and the type probability is updated through the intensity function.Secondly,the Gibbs sampling,BIC criterion and Gibbs-GLMB algorithm are introduced.Finally,the type probability is brought into the GLMB algorithm to form the typed GLMB algorithm.(2)A tracking estimation method of mixed targets is proposed according to the GLMB-based type probability algorithm.Firstly,state estimation is carried out on the mixed targets with structural relations,and the dynamic modeling of the mixed targets is carried out based on the correlation graph theory.Then the motion model of the mixed target is established,which can let the point target and the extended target perform CV motion,and the group target perform CT motion.Finally,the mixed target is tracked according to the type probability algorithm,and the performance of the algorithm is measured according to the OSPA algorithm.(3)A GLMB-based target tracking estimation algorithm is proposed for longperiod and large batches of mixed targets.First,the long-period root mean square error,maneuver index,loss rate,and large and small process noise will be analyzed.Then,a system model of long-period targets is established,and the tracking performance of two types of targets,namely uniform turning motion and uniform acceleration motion,in long periods are analyzed separately.Finally,we will analyzed the problems that will occur when the target tracking is gradually transitioned from small batches to large batches,and we will use the time-for-space method to track large batches of targets,and realize it through simulation...