Research On Multi-target Passive Tracking Based On Feature Aided Data Association Technology

Passive sonar detection technology as always been one of the main methods for underwater target detection because of its strong concealment and excellent detection performance.Multitarget passive tracking is an important part of the passive sonar detection system,and it is an important technical guarantee to continuously track the target and obtain the target parameter information,so the research of passive sonar multi-target tracking algorithm has a very important significance.Firstly,this paper studies two aspects of multi-target bearing only passive tracking technology: bearing only passive tracking filter and bearing only data association.It includes modified polar coordinates bearing-only passive tracking filter,nearest neighbor standard filter,strongest neighborhood filter and joint probability data association algorithm.The target tracking performance of the modified polar coordinate extended Kalman filter is simulated and analyzed in the clutter free background.In the background of clutter,the simulation analyzes the performance of single target passive tracking based on the nearest neighbor standard filter and multi-target passive tracking based on the joint probability data association.Then,in view of the conventional multi-target bearing-only passive tracking technology,the tracking performance deteriorates when the multi-target bearing crosses and approaches,the multi-target passive tracking algorithm based on feature-aided correlation is studied.It includes setting up multi-target tracking model,target bearing-frequency extraction algorithm based on Dirichlet process Gaussian mixture model,multi-target passive tracking algorithm based on feature aided association.Comparison simulation experiment is carried out: On the basis of acquiring bearing-history based on beamforming,multi-target passive tracking algorithm based on feature-aided association and multi-target passive tracking algorithm based on joint probabilistic data association are simulated and compared for cross target signal level tracking.Monte Carlo simulation is used to analyze the error association probability,the number of track interruptions and the tracking accuracy of bearing and frequency information.Based on the sea trial data,the tracking performance of the weak target,the tracking performance of the weak target and the tracking filtering performance under the condition of close trajectory and interference are verified.Finally,the realization of multi-objective passive automatic tracking algorithm module is studied,including algorithm module framework and algorithm module function test.The module framework of the algorithm mainly studies the overall algorithm module framework and process,which combines multi-target passive tracking algorithm based on bearing-only strongest neighborhood association,multi-target passive tracking algorithm based on featureaided association,and trajectory termination algorithm based on successful proportion of trajectory detection.In the function test part of algorithm module,the real-time performance,tracking accuracy and functional integrity of algorithm module are analyzed by using simulation and test data.