Research On Track-Before-Detect Algorithms For Weak Target Based On Particle Filter

In modern warfare,it is crucial to detect the enemy threat and respond as early as possible.With the development of modern technology,a lot of low echo energy targets,such as stealth aircraft and cruise missile appear more frequently in the warfare.The traditional detect before track method(DBT)is no longer suitable for track and detect weak target under the low SNR circumstances.Track before detect algorithm(TBD)performs well in track and detect weak target under the low SNR circumstance.This algorithm has gained more and more attention.This thesis focuses on the research of track-before-detect algorithms for weak target based on particle filter(PF-TBD).PF-TBD algorithm has its unique advantages in tracking and detecting weak target in non-linear,non-Gauss system,but the standard PF algorithm itself has its own problems and limitations,which to some extent affect the performance of PF-TBD algorithm.This thesis focuses on the research on improved PF-TBD algorithm form two different angels,which are the selection of importance probability density function and resampling improvement.Firstly,based on the traditional standard PF algorithm,unscented particle filter(UPF)algorithm and Markov chain Monte-Carlo(MCMC)moving algorithm based on MH(Metropolis Hastings)criterion are introduced.Form these two aspects unscented particle filter with MCMC moving algorithm is proposed.Aiming at the problem of ill-timeliness caused by the complexity of the algorithm,by synthesizing the advantages of standard PF algorithm,the improved algorithm is proposed,and simulation experiments is conducted to verify the effectiveness of the improved algorithm.Furthermore,in consideration of the advantages of UPF algorithm,UPF-TBD algorithm is proposed.Because the traditional PF-TBD algorithm's resampling step can cause particle diversity loss and even lead to the problem of particle exhausted,genetic algorithm is used to improve the algorithm.In consideration of the advantages of MCMC algorithm,genetic resampling algorithm based on MH mutation is proposed.This algorithm is used as the alternative algorithm in the UPF-TBD algorithm resampling step,which is an improved algorithm.Simulating the constant velocity model to verify the effectiveness of the improved algorithm.Finally simulating the weak target detection and trace under infrared circumstance model to further verify the effectiveness of the improved algorithm.