Research On Underwater Target Tracking Algorithm Based On Particle Filter

As marine development and underwater operations have become more and more concerned by the world,underwater target tracking is an indispensable part of it,which is becoming a research hotspot that people pay more and more attention.Underwater target tracking is a complex process,it quickly estimates the state of a moving target as close as possible to the real state by the measurements collected by the underwater sensors.The problem of estimating the states of targets is a nonlinear filtering problem in essence,and particle filter has the excellent property that it is not constrained by linear Gaussian conditions,which makes it an effective method to solve the state estimating problem in target tracking.This paper studies the problems related to underwater target tracking based on particle filter.To deal with particle degradation and particle impoverishment which lead to not high tracking accuracy,particle filter is improved by studying the importance density function and the resampling step.The main work of the paper is as follows:First,in this paper,an improved particle filter algorithm is proposed.The square root cubature Kalman filter is used to generate the proposal distribution.The self-adaptive artificial fish swarm algorithm is employed to optimize the particles.The prey behaviour and swarm behaviour of fish swarm make the particles closer to the true posterior distribution,solving the particle degradation and impoverishment problem.Moreover,a sensor selection scheme is designed for underwater wireless sensor networks,which reduces energy consumption of the networks by exactly waking up four sensor nodes at each time,while preserving tracking accuracy.At the same time,aiming at the distributed information fusion problem of multi-sensor nodes,a fusion method based on local estimates' similarity is proposed to obtain better tracking results.The simulation results demonstrate that the proposed method has superior tracking performance.Secondly,aiming at the problems of standard particle filter,this paper proposes another improved particle filter algorithm by combining particle filter with artificial bee colony algorithm,which is applied to target tracking via underwater wireless sensor networks.In the proposed method,the square root cubature Kalman filter is used to generate the proposal distribution,and the artificial bee colony algorithm is employed to optimize the particles before resampling.Each particle is regarded as a honey source,the work that the bees perform searching local optimal food source is the process of optimizing particles,which increases the number of effective particles and the diversity of particles.In addition,the linear minimum variance criterion is utilized in the distributed fusion structure of underwater wireless sensor networks to fuse local estimates together.The simulation results suggest that the proposed algorithm outperforms other classical algorithms in tracking accuracy and efficiency.