Research On Multi-target Particle Filter Tracking Algorithm Before Detection Based On Particle Swarm Optimization

In the multi-radar multi-target joint detection and tracking problem,it is the main task and hot spot for researchers at home and abroad to find targets correctly and improve their tracking accuracy in time.Particle filter based track-before-detect algorithm is a classic track-before-detect algorithm.This algorithm relies on a set of particles with weights to approximate the posterior probability density function of the target state.It is not limited by nonlinear and non-Gaussian conditions.Wide adaptability is an important direction in the multi-radar and multi-target joint detection and tracking algorithm,which has important theoretical research significance and national defense military value.When detecting and tracking multiple targets with large differences in signal-to-noise ratio,the traditional track-before-detect algorithm based on particle filtering easily replicates a large number of high-weight particles when performing particle swarm resampling,resulting in only detecting targets with high signal strength,missing targets with relatively weak signal strength may cause missed detection,or a large number of low-weight particles may be sampled,leading to false targets.In order to improve the correct target detection rate in the multi-radar multi-target detection and tracking problem,this paper is based on the structure of the double-layer particle filter algorithm,a multi-radar multi-target track-before-detect algorithm based on particle filter is studied.The main work is as follows:1.Firstly,the basic principles of the particle filter based track-before-detect algorithm,target motion model and radar observation model are introduced,and the specific implementation steps of the multi-radar single target particle filter based track-before-detect algorithm are described.2.A multi-radar multi-target detection tracking algorithm(IGPF-TBD)based on adaptive genetic operation is proposed to improve the problem of missed target detection when the target signal strength differs greatly.The algorithm is inspired by genetic algorithms.It generates new particles through adaptive genetic operations,improves the diversity of particle distribution.The tournament selection method is used,and the relative value of particle weights is used as the criterion for selecting particles.The sampling probability of particles with higher weights is increased to avoid a large number of high-weight particles from being copied,thereby improving the missed detection of targets.The simulation results show that compared with the traditional multi-radar multi-target particle filter based track-before-detect algorithm,the algorithm can effectively detect targets with small signal-to-noise ratio when the target signal-to-noise ratios are significantly different.3.A multi-radar multi-target particle filter track-before-detect algorithm based on particle swarm fusion(PSFPF-TBD)is proposed to improve the target tracking accuracy of the multi-radar multi-target track-before-detect algorithm.The algorithm designs a particle swarm fusion idea,which is used for the organic fusion of particle swarms for newly detected targets and particle swarms for discovered targets.By reasonably retaining the dominant particles of the two types of particle swarms,the target tracking particle swarm status is improved and target tracking precision.Simulation results show that the algorithm can effectively improve the target tracking accuracy compared with the traditional multi-radar multi-target particle filter track-before-detect algorithm.4.This paper proposes a hybrid multi-radar multi-target track-before-detect algorithm(HMPF-TBD),which is used to improve the target miss detection problem and improve the tracking accuracy in multi-radar multi-target track-before-detect problem.Based on the double-layer particle filter structure,the algorithm organically fuses the algorithm structures of IGPF-TBD and PSPPF-TBD,and combines the advantages of the two algorithms.Simulation results show that the hybrid algorithm can improve the target detection rate and target tracking precision.