Formation Target Tracking Technology Based On Random Finite Set And Neural Network

With the rapid development of military modernization in various countries,camouflage and stealth of targets are greatly strengthened.The detection of formation targets(such as aircraft formation,missile formation,surface ship formation,UUV and AUV formation)by sensors(radar,sonar,etc.)is facing great challenges.At the same time,due to the uncertainty of the number of formation targets and the influence of complex environment clutter,there are still great technical difficulties in formation target tracking.This paper focuses on the formation target tracking algorithm based on the combination of random finite set and neural network.Firstly,the basis of formation target tracking based on stochastic finite set is studied,and the measurement observation model based on passive double base stations is studied.Then the related theoretical knowledge of stochastic finite set is studied,including the establishment of stochastic finite set model,finite set statistical theory and common random finite set distribution.At the same time,Gaussian realization of probability hypothesis density filter and multi-target evaluation index of tracking algorithm are studied.Secondly,several common motion models of target tracking and the derivation formulas of some common filtering algorithms are studied,and the filtering effects of the basic filtering methods studied in this chapter under the same motion model and different observation models are simulated and compared.Finally,the formation target tracking algorithm based on the combination of random finite set and neural network is studied,and the common formation is studied.In order to solve the problem of formation target tracking,EKF / UKF-GM-PHD algorithm based on GM-PHD is studied.Aiming at the problem that the GM-PHD model can not accurately estimate the number of targets in low signal-to-noise ratio and complex target motion.The EKF / UKF-GM-CPHD algorithm based on GM-CPHD is studied.In order to improve the tracking accuracy of formation target,the tightly coupled tracking model of Elman neural network and UKF algorithm is studied,and the ELM-UKF-GM-CPHD algorithm is studied,and the effectiveness of the above algorithm is verified by the algorithm simulation under different formation movements.