Research On The Underwater Multi-Target Passive Tracking Technologies Based On The Random Finite Set

Underwater multi-target tracking technology has always been a hot research topic in the field of sonar,with an extensive application value in the military defense and ocean exploitation.With the increasing number of underwater targets and the improvement of the invisibility,multitarget tracking performance is affected by low signal-to-noise ratio(SNR),target crossing and closely spacing scenario,which can not satisfy the tracking accuracy and stability demand of the passive sonar system.In order to accommodate the aforementioned scenario,this dissertation focus on improving passive sonar tracking performance under the low SNR and dense targets environment.Furthermore,this dissertation solves the problems including filtering,track maintain and the multiple platform data problem based on the random finite set(RFS)theory.Considering the unknown and time-varying number of targets in the surveillance region,we introduce the random finite set theory to capture the target presence/absence in the surveillance region as well as its kinematic state,and form the multi-target motion model accounting for the target birth and death.Furthermore,we study several multi-target Bayesian filtering algorithms and the evaluation criteria of tracking performance,which form the basis of this dissertation.In order to improve the direction-of-arrival(DOA)tracking performance in the low SNR environment,we propose a superpositional approximate multi-Bernoulli tracking algorithm using the passive sensor array measurement.First,we derive the multi-target likelihood function based on the statistical model of the passive sensor array measurements.Under the multi-Bernoulli RFS filtering framework,we derive a computational tractable multi-Bernoulli filtering equations by using superpositional approximation and achieve the joint estimation of the number of targets as well as its DOA.The proposed method directly conducts on the sensor array measurement,which preserves more information and allows better DOA tracking performance in the low SNR environment.The simulation results verify the effectiveness of the proposed method.In the scenario of trajectory crossing,the DOA of two moving targets are closely spaced and overlapped.Due to the resolution of the sensor array,the beamforming result presents a single peak,and forms a merged measurement of two targets,which can cause the incomplete and inconsistent trajectories.In order to solve this problem,we develop a superpositional approximate generalized labeled multi-Bernoulli tracking algorithm using the passive sensor array measurement.The proposed method directly conducts on the sensor array measurement,which avoids the influence of merged measurements.Furthermore,the proposed method combines the track management and the Bayesian filtering procedure.The track management procedure can form and maintain the tracks of the crossing targets.Simulation and sea trial results verify the effectiveness of the proposed method.Considering the multi-platform multi-target data association problem in the scenario of unknown and time-varying number of targets,we propose a multiple platform multi-target passive tracking algorithm based on cross entropy data association.The proposed method combines the data association and the filtering procedure,and constructs the probability distribution on the space of all multi-platform data association.It gives the possible circumstances of the measurement-to-track association as well as the targets birth/death by using randomly sampling,and improves the multi-target tracking performance in the scenario of unknown and time-varying number of targets.The simulation results verify the effectiveness of the proposed method the show its value in applications.