Research On Multi Maneuvering Target Tracking Algorithm Based On Interacting Multiple Models Fast Data Association

As a research hotspot in recent decades,the role of maneuvering target tracking technology is growing in military and civil areas.With the development of maneuvering target tracking technology,various kinds of model,filtering and data association algorithms have been proposed,improving the performance of algorithm greatly.In addition,the improvement of target maneuvering performance also proposed more and more severe test to target tracking system.Therefore,the study of maneuvering target tracking algorithm has great significance both in theory and in practice.The paper introduced the composition of target tracking,basic principles,as well as some typical mathematical model,including the CV model,CA model,singer model,current statistical model,CT model,and summarizing the characteristics of each model.Second,filtering algorithm in target tracking was introduced,including Kalman Filter,Extended Kalman Filter(EKF),Unscented Kalman Filter(UKF),and compared filtering effects of the three algorithms through simulation.Then,according to the problem of data association,studied the nearest-neighbor data Association(NN),probability data Association(PDA)and the joint probability data Association(JPDA).JPDA is recognized as the classic algorithm of multi-target data association,but the algorithm needs to search for all possible solutions,and the exponential growth effect leads to excessive computation.Pointing at this disadvantage in the JPDA algorithm,fast data association(FDA)algorithm is proposed in this paper to solve the problem.As the focus on content,this paper analyzes the basic principle of interacting multiple model algorithm(IMM)in detail.Aiming at the problem of maneuvering target tracking in clutter,the IMM and FDA algorithm are combined to solve the uncertainty problem of the target tracking and target motion state.In order to solve the problem of target miss in tracking process,this paper applies the extended association gate to this algorithm.When the effective measurement can't be detected in the correlation gate,it searches by the way of gradually expanding the correlation gate,which can reduce the loss rate of the maneuvering target tracking.Meanwhile,in order to reduce the computational complexity of the new algorithm,the predefined sampling interval method is used to adaptively adjust the tracking sampling period and balance the target tracking accuracy and tracking system load.