Research On Target Tracking Technology Of Distributed Networked Radar Based On GM/SMC-PHD Filtering

The probability hypothesis density(PHD)filter based on Random Finite Set theory is a kind of Target tracking methods gained widespread attention and development in recent years,especially for multiple targets tracking(MTT),which integrates the start,maintain and end of target tracks,avoiding some serious problems such as combination explosion and associated errors of traditional MMT ways,and it can additionally work at more complex tracking environment effectively.In order to improve accuracy and tracking stability,a way of collaborative tracking using networked radar systems is a necessary choice in addition to the development and optimization of tracking methods themselves,such as the distributed networked radar system which is paid much attention particularly because of its excellent robustness and low cost.The thesis focuses on the research of tracking ways for multiple targets with a lot of motion models and distributed collaborative tracking ways based on classical PHD filtering methods.The specific content mainly includes the following three aspects:1.The summerization of the development thread of radar target tracking technologies and the research status at home and abroad,and the introdution in detail of basic related theories and methods required by this thesis,including the single target tracking,MTT and maneuvering tracking technologies and the multi-sensor joint tracking technology.2.The research on tracking methods for multiple maneuvering targets based on the PHD filtering technology.Different improved ways are presented to be suitable for maneuvering targets environment aiming at two mainstream PHD filtering methods,Gaussian mixture probability hypothesis density(GM-PHD)and sequential Monte Carlo probability hypothesis density(SMC-PHD)filter: a multiple model GM-PHD tracking way based on the posteriori probability of current moment motion models with the combination of the GM-PHD filtering algorithm and multiple model(MM)theory;An adaptive weight extension particles multiple model sequential Monte Carlo probability hypothesis density tracking way based on the Markov transition probability matrix combining the SMC-PHD filtering algorithm and MM and interacting multiple model theory.Then the simulation software is used with application scenes set up separately for such two methods for experimental analysises to prove their effectiveness.3.The research on tracking methods of the distributed radar system based on the PHD filtering technology.In view of low anti-interference,poor flexibility and relatively weak tracking ability of a single radar,a distributed GM-PHD tracking way based on the K-means clustering technology and a distributed SMC-PHD tracking way based on the K-means clustering technology are proposed in the principle of the distributed radar system.Similarly,the simulation software is used with application scenes set up separately for such two methods for experimental analysises to prove their effectiveness.