Research On Maneuvering Multi-target Tracking Technology Based On δ-GLMB Filter

With the continuous development of sensor technology,multi-target tracking systems have become an important component of modern military attack and defense systems.They have broad application prospects in military fields such as aerial reconnaissance and early warning,missile defense,surveillance of the battlefield,as well as civilian fields such as autonomous driving,machine vision,and medical image processing.In recent years,multitarget tracking technology based on random finite set has gradually become a research hotspot because it solves the combinatorial explosion problem caused by the increase of the number of targets in the traditional data association technology.Taking maneuvering targets as the research object,this paper improves the δ-generalized labeled multi-Bernoulli filter(δ-GLMB)in the tag random finite set from the aspects of maneuvering multi-target tracking,nonlinear measurement processing,Doppler radar measurement application,target adaptive regeneration,etc.The specific research contents are as follows:(1)In order to improve the performance of the δ-GLMB filter in maneuvering multiple targets tracking,a new multi model Gaussian linear δ-GLMB algorithm is derived based on the interactive multi model framework in single target tracking,which solves the problem of filtering divergence caused by the lack of necessary information interaction when system parameters change in static multi model algorithms.This algorithm uses multiple motion models,each of which is transferred through a Markov probability transfer matrix,effectively improving the δ-GLMB filter’s adaptability to complex maneuvering scenarios.(2)A multiple model δ-GLMB filter using decorrelation unbiased measurement transformation and fuzzy algorithm improvement is proposed to address the issues of high computational complexity,large tracking error,and poor robustness to clutter changes in maneuvering multi-target δ-GLMB algorithms under nonlinear measurements.This algorithm uses decorrelation unbiased measurement transformation Kalman algorithm to replace traditional nonlinear processing methods such as extension and volume Kalman to reduce computational complexity,and proposes a joint gate clutter filtering strategy to eliminate false alarm measurements.Finally,an improved fuzzy algorithm is introduced to adaptively change the process noise of the operation model,further increasing the tracking accuracy of the filter.(3)Aiming at the defect that the δ-GLMB filter can only detect new targets at fixed positions due to insufficient measurement information in Doppler radar scene,an interactive multiple model adaptive new measurement conversion sequential δ-GLMB filter for Doppler radar is proposed.The algorithm introduces the Doppler radial velocity information in the Doppler radar,carries out measurement conversion,decorrelation and sequential filtering for the position measurement information and radial velocity information in turn,and uses measurement drive to determine the state distribution of the new target,so as to realize the adaptive regeneration of the target in the whole detection area.The simulation results show that the proposed algorithm effectively utilizes Doppler measurement information and can improve the tracking accuracy of maneuvering multiple targets while adaptively detecting new targets.