Research On Multi-target Tracking Technology For Ammunition Fragments

In the military field,the method of tracking ammunition fragments can be used to verify the accuracy of the researched large-array radar target tracking results.However,the use of live ammunition blasting often faces the problem of high cost.Therefore,it is necessary to establish a simulation model through computer simulation,and establish the movement model of the fragment through the movement reflected by the simulation model,so as to realize the tracking of the scattered fragment target.In terms of fragment target tracking,due to the characteristics of high density and large number,it is extremely difficult to track them.Therefore,the research on the tracking technology for fragment targets has important practical significance and military value.This paper focuses on the research of multi-target tracking technology for fragmented targets.The research directions include the following two points:(1)Finite element simulation of the fragment flying process,which involves the basic motion theory of fragments and fluid-solid coupling mechanics in the end effect science,including finite element simulation model modeling,material parameter setting and deduction of fragment-related data.(2)Tracking multiple fragmented targets,which involves the related content of the multitarget tracking algorithm,including the traditional multi-target tracking algorithm based on data association and the multi-target tracking/extended target tracking algorithm based on random finite set.This paper focuses on the above two research directions,and the main content of the research includes the following three aspects:(1)The finite element simulation of the process of generating scattered fragments by the blasting of the ammunition is carried out.The fluid-solid coupling mechanics method is used to establish a finite element model combined with the basic theory of fragment movement,and the simulation results are obtained by calculation.According to the physical phenomena shown by the simulation results,the correctness of the simulation model is preliminarily verified;the simulation data obtained from the model is combined with the method of data deduction,combined with the relevant theoretical formulas of end-effects,and the angle,speed,etc.of the simulated fragments and verify the correctness of the model.(2)According to the results of the simulation model,the fragment motion model is established,and the traditional data association-based multi-target tracking algorithm is used to track the fragment target.Aiming at the performance deficiencies of the traditional data association algorithm—JPDA algorithm after the number of targets increases,this paper is based on the aggregation which propose an improved GMM-EM algorithm,and compared it with other improved JPDA algorithms such as IJPDA and FCM through related simulation experiments.While proving the feasibility of the algorithm,it also demonstrated its large number of targets.The superiority of tracking.(3)Due to the shortcomings in the clustering of the GMM-EM algorithm itself,the multitarget tracking algorithm based on random finite set is studied,and the Label Multi-Bernoulli(LMB)algorithm is used to track the fragmented target,and it is verified that it is compared with The GMM-EM algorithm has more advantages;at the same time,in view of the long operation time of the LMB algorithm,the target group composed of fragments is regarded as the extended target,and the GGIW-LMB algorithm is used to track the LMB extended target,and it is verified that it is not While reducing the accuracy of the original LMB algorithm,it also reduces the computing time;finally,in view of the potential underestimation problem in the GGIW-LMB algorithm,this article improves the measurement partitioning steps in the algorithm,and introduces the measurement partition based on state prediction methods,the TP-GGIW-LMB algorithm is proposed,and simulation experiments prove that it can solve the problem of underestimation in the GGIWLMB algorithm while maintaining the rapidity of the GGIW-LMB algorithm.