Joint Detection And State Estimation Of Dim Targets In Pulse Doppler Radar

In the face of the threat of anti-radiation missiles,low-altitude and ultra-lowaltitude penetration and stealth technology in the development of radar,dim target detection and state estimation under clutter and noise background are difficult problems to be solved in modern radar technology.On the one hand,the traditional detect before track(DBT)method is easy to lose the dim target information in the detection link,on the other hand,it is easy to have a computational explosion in the tracking link due to the influence of false alarm and multi-target factors.In view of the defects of DBT method,based on random finite set(RFS)to estimate the number and state of targets at the same time in Bayesian framework,this paper mainly studies the joint detection and defuzzification of multiple targets of pulse Doppler radar under low signal to clutter ratio(SCR),and the joint detection and state estimation of single target and multiple targets of pulse Doppler radar under low signal-to-noise ratio(SNR).The main work is summarized as follows:Firstly,the basic description symbols of RFS and several important RFS distributions are introduced.The RFS are compared with random vectors to reflect the advantages of RFS in the field of multi-target tracking.In addition,the general multiobjective Bayesian filter is analyzed to provide a theoretical framework for the subsequent RFS filter.Then,the joint multi-target detection and range Doppler ambiguity resolution of pulse Doppler radar under low SCR are studied.Based on the detector output model,the analytical formula of cardinality balanced multi-target multi-Bernoulli(CBMe MBer)filter is given.The defuzzification process is modeled into the observation equation,and the particle filter(PF)based on adaptive new density is used to estimate the number of multiple targets and defuzzify at the same time.Experiments have verified that the algorithm can effectively estimate the number of targets and the real state under high clutter density.Compared with the particle based cardinalized probability hypothesis density(CPHD)filter,the filter has better detection and estimation performance and higher operation efficiency.Finally,the joint detection and state estimation of single target and multi-target of Doppler radar under low SNR are studied respectively.Aiming at the problem of single target detection and estimation,the analytical formula of Bernoulli filter based on intensity measurement model is given.The target state vector is extracted and modeled at the signal level.The single target joint detection and state estimation are realized through particle filter.Simulation experiments show that the algorithm has excellent detection and estimation performance under low SNR.Aiming at the problem of multitarget detection and estimation,considering the superposition of measurements on the basis of single target,and combining the advantages of simple structure of multiBernoulli(MBR)filter and accurate potential estimation of CPHD filter,a hybrid MBRCPHD filter based on superposition sensor model is proposed,and multi-target joint detection and state estimation are realized based on auxiliary particle filter(APF).The simulation results show that the filter can estimate the number and state of targets more accurately than the MBR filter based on superimposed sensors,and maintain lower computational complexity.