Researchs On Multi-feature Based Multiple Hypothesis Tracking Algorithm And Track Fusion Method

In recent years,the electromagnetic environment has become increasingly complex,and armaments have been upraded in many contries.Various supersonic,highly maneuverable,and camouflaged stealth devices are applied in modern battlefields.In this case,radar systems face severe problems such as low detection probability,high false alarm rate,the discontinuity of the estimated trajectory,and poor tracking performance under strong clutter,and there is an urgent need to improve the tracking performance of single radar or adopt the radar network to solve these problems.Therefore,this thesis proposes an improved multiple hypothesis tracking algorithm for realistic dense clutter scenarios.Compared with the original algorithm,the proposed algorithm can improve the accuracy of the data association procedure and the precision of the estimated trajectory.Considering the discontinuity of the estimated trajectory and the poor accuracy of the tracking algorithm,this thesis,based on the requirements of projects,deeply studies the track fusion algorithms for the radar network from multi aspects,and serveral simulation systems are developed.The main work and innovations of this thesis are as follows.(1)This thesis introduces several commonly used filtering algorithms in engineering,including the Kalman filter(KF)for linear scenarios,the extended Kalman filter(EKF)for non-linear scenarios,the converted measurement Kalman filter(CMKF)and the cubature Kalman filter(CKF).Through the theoretical analyses and simulation experiments,the performaces and the scope of applications for these algorithms are summarized.(2)The data association method is used to achieve the correct associating of measurements and targets,which is a prerequisite for subsequent filtering estimation and track fusion.Therefore,this thesis first introduces the classical nearest neighbour(NN)algorithm and the joint probabilistic data association(JPDA)algorithm,and designs simulation experiments to verify their tracking performances under clutter.Second,we focus on the basics of multiple hypothesis tracking(MHT)algorithms,including hypothesis event generation,hypothesis probability calculation and several hypothesis reduction techniques.For the modern battlefield environment with dense clutter,the data association is carried out jointly using doppler features,amplitude features and original position features of the target,and a multifeature based multiple hypothesis tracking algorithm is proposed.Simulation results imply that the proposed algorithm can effectively improve the accuracy of the data association method.At the same time,through the charactristics of delayed decision in the multiple hypothesis tracking algorithm,we apply the smoother to the track update procedure,and,therefore,the accuracy of the output track is improved.(3)Based on the requirements of the projects,this thesis focuses on the track fusion algorithm in radar network systems.First,time and spatial alignment techniques are briefly introduced to enable the conversion of every local radar data to a common coordinate system,which can then be involved in the subsequent fusion.Second,we mainly analyze simple convex combination algorithms,covariance Intersection algorithms for synchronous network systems,and sequential fusion algorithms for asynchronous tracking systems.Finally,we focus on the study of track fusion algorithms through the numerical simulation experiments,processing of real data,and the development of the network simulation system,which provides a useful basic research and an experimental platform for the radar network system in actual engineering.