Research On Underwater Target Tracking Method Based On Particle Filter And Joint Probabilistic Data Association

In recent years,with the development of underwater acoustic technology,underwater target tracking technology has received more and more attention.It plays a vital role in making full use of our marine resources and safeguarding our marine rights and interests.It has wide application value and important strategic significance in wartime surveillance,sonar detection,underwater weapon confrontation,and the development of marine resources.Aiming at the two hot points of underwater target tracking,particle filter algorithm and joint probabilistic data association algorithm,theoretical research and simulation research are carried out,with emphasis on improving the real-time calculation and tracking error.Firstly,the basic filtering theory is expounded in this dissertation.The common system models are given,and the basic Bayesian filtering estimation theory and representative algorithms,Kalman filtering algorithm and extended Kalman filtering algorithm,are studied.Because the underwater target and detection environment are more complex and changeable,and Kalman filter algorithm can not deal with the non-linear and non-Gaussian system better,then the particle filter algorithm is studied in depth,and the principle and process of particle filter algorithm are elaborated in detail.Then,an improved particle filter algorithm based on density function is proposed to deal with the particle degeneration easily occurring in particle filter algorithm.By analyzing the performance of the five filtering algorithms from the aspects of estimation state error and algorithm running time,the extended Kalman filter algorithm is selected as the density function of the nonlinear system to guide sampling.In order to reduce the amount of calculation,the Kalman filter algorithm is selected as the density function of the linear system,and the line is given.The judgment basis of sex system and non-linear system.The simulation results of linear,non-linear and synthetic cases show that the improved algorithm improves the estimation accuracy and reduces the operation time of the algorithm compared with the extended Kalman particle filter algorithm,which satisfies the real-time requirement of underwater target tracking.Finally,an improved method based on joint probabilistic data association algorithm is proposed to solve the problem of track cross-merging in underwater multi-target tracking.By choosing the feasible event with the greatest probability of related events,the influence of other points falling in the common area of tracking gate is reduced,and the merging problem between tracks is effectively reduced.Through the vertical moving target,parallel moving target and cross moving target,three groups of simulation experiments verify that the improved algorithm can effectively prevent the cross-merging of trajectories and reduce the tracking error.