Research On Target Tracking Approach Of The Mission Based Unmaned Underwater Vechile

Unmanned underwater vehicle(UUV)has received wide attention in military and civilian aspects,using UUV to surveillance the underwater environment has become an important method,multiple target tracking of UUV provides the essentially target state information to implement ASW mission.For sovling the multiple maneuvering target tracking problem this paper study on the mathematical model of underwater motioning target,the single target tracking method based on Bayes filter,the multiple target tacking and the performance evaluate method based on Random set,the underwater multiple maneuvering target tracking method and so on.(1)Based on the characteristic of underwater target,devided the mathematical model of non-maneuvering target and maneuvering,including constant velocity model,accelerate velocity model,constant turining angle model and “current” statistics model,and the distribution time equations of each model are presented.Based on the theory of Bayes estimate,the principle of Kalman filter,extended kalman filter,unscented kalman filter and particle filter are analysing,and the recursion equations are presented.Finally,by designing the simulation experient of single target tracking,the KF effective is proved in Gaussian noise linear system,and the EKF,UKF,PF in Gaussian noise non-linear system,and their performance are contrasted.(2)Analysing the feasibility of the multiple target tracking method based on random set,then based on the random set establish the multiple target state model and target observation model used by Bayes recursion estimate,discussing the relation between the Bayes filter and Probability Hypothesis Density filter,and the PHD filter recursion equations are presented,then study on the evaluate method of multiple target tracking performace based on PHD filter,this has established the foundation for the future research.(3)Because PHD filter can not be achieved in the enginerring,two kinds of achieved way for the PHD are presented: the Particle PHD filter and the Gaussian Mixture PHD filter,and design the multiple target tracking simulation experient for each filter,proved the effective of the filters.But considering P-PHD filter has the weak of the large computing and the low filter accuracy in the non-linear system,this paper uses the EKF and UKF to predict and update the Gaussian component in GMPHD filter,the EK-GMPHD filter and the UK-GMPHD filter are presented.Finally,by designing the multiple target tracking simulation experient,declared the performance of UK-GMPHD filter is better than EK-GMPHD filter.(4)To accomplish the ASW mission with UUV,for the problem of multiple maneuvering target tracking,UUVdetect the information of target by carrying sonar,achieved the observation information of target.This paper combines the multiple model method and UK-GMPHD filter,the MM-UKGMPHD filter is presented.By analysing the task of ASW mission,serveral kinds of target motion form are designed,UUV is assumed to sailing in ideal scenario,the multiple maneuvering target tracking simulation experient proved the UUV can complete the ASW mission.