Research On Maneuvering Target Tracking Algorithms Based On Multiple Passive Sensors

With the rapid development of the aerospace technology, both the speed andmaneuverability of all flights are becoming higher and higher. These factors make themaneuvering target tracking problem, especially multiple maneuvering targets trackingin the clutter environment be of important theoretical significance and practical value.Passive sensor has a wide application in modern surveillance system and weaponsystem due to its good stealth and strong anti-interference adaptability.In this thesis, we mainly study the problems of maneuvering target tracking formultiple passive sensors. The main contents include maneuvering target tracking models,maneuvering target tracking algorithms under Gaussian noise and non-Gaussian noise,and multiple maneuvering target tracking algorithm based on the data associationtechnique and the random set theory. Main contributions are as follows:Firstly, an improved interactive multiple model algorithm based on the Viterbi isproposed. On the basis of the Viterbi principle, an interactive multi-model Viterbialgorithm is proposed by combining the interactive multiple model method and theViterbi technique, which can effectively improve the performance for a particular targettracking. Furthermore, a switching Viterbi interactive multiple model algorithm isproposed by introducing a maneuvering detection mechanism before the model switch,allowing the different Viterbi parameters to better describe the state of target motion.Secondly, a new multiple maneuvering targets tracking algorithm is presented,based on the study of Markov chain Monte Carlo sampling methods. In this algorithm,data association is achieved by sampling the associated events, is different from theconventional joint probabilistic data association algorithm enumerating all feasibleevents. The proposed algorithm greatly reduces the computational complexity andsimplifies the problem of multiple target tracking.Finally, for the target appearing, disappearing, spawning problems, multiplemaneuvering targets tracking algorithm based on the Finite Set Statistics (FISST) theoryis studied. An improved multiple model probability hypothesis density algorithm ispresented, which improves the multi-target tracking performance when model transitionprobability is too small. In addition, we further study the optimal cross-entropy theoryand combine it with the probability hypothesis density method. A novel trackmaintenance algorithm is proposed, which can effectively improve the track maintenance performance.